# Changelog
Source: https://docs.aident.ai/changelog

See Aident AI's latest updates for our Aident Playbook Editor

<Update label="May 2026" description="5.29.2026 | v0.2.11b">
  ## Onboarding, Vault, and Cleaner Connections

  **New Features**

  * **Launched** guided onboarding so new users can tell Aident what they want to build before landing in the app
  * **Added** secure vault-backed credential flows for richer CLI and integration connections
  * **Introduced** hosted CLI groundwork, including sandbox runtime support and the first seeded CLI integrations
  * **Added** trial checkout support so eligible users can start from the billing flow without extra help

  **Improvements**

  * **Improved** mobile playbook, billing, usage, and capabilities layouts so core flows stay usable on smaller screens
  * **Enhanced** Composio recovery to retry omitted actions, recover deprecated actions, and clean snapshots around executable tools
  * **Optimized** onboarding migrations, route timeouts, and app-shell billing navigation for fewer stuck or stale states

  <span>Bug Fixes</span>

  * **Fixed** integration reconnects not always resuming cleanly
  * **Fixed** connected integrations not always appearing under Mine
  * **Fixed** integration retry behavior leaking outside the chat card that requested it
  * **Fixed** several mobile playbook alignment issues around chat turns, output summaries, trigger toggles, and delete actions
</Update>

<Update label="May 2026" description="5.25.2026 | v0.2.10">
  ## Published Outputs & Capability Polish

  **New Features**

  * **Launched** shareable published HTML outputs so generated pages can open from a stable public link
  * **Added** native API credential connections, including support for body-auth integrations and a Tushare API fixture
  * **Introduced** new capabilities audit and vault surfaces for reviewing connected actions and credentials
  * **Built** inline reply input for playbook runs waiting on a user answer

  **Improvements**

  * **Improved** capability discovery with better environment visibility, native API search, and action alias resolution
  * **Enhanced** mobile playbooks with Run Trace access, visible run controls, and better chat spacing
  * **Added** design system icon updates and clearer published result handling across playbook outputs

  <span>Bug Fixes</span>

  * **Fixed** stream timeout fallbacks and executor finalization so runs recover more reliably
  * **Fixed** stale or finalized sessions that could leave playbook runs looking stuck
  * **Fixed** OpenAPI integration payload handling and server-side creation paths
  * **Fixed** maintenance banner text customization and layout accounting
</Update>

<Update label="May 2026" description="5.16.2026 | v0.2.9a">
  ## Public Catalog & Realtime Playbooks

  **New Features**

  * **Launched** public integrations and skills catalog pages so people can browse what Aident supports before signing in
  * **Added** an assistant route skeleton with sidebar loading states for a smoother first render
  * **Introduced** capability index events and local index population so search results stay fresher
  * **Improved** MCP OAuth resilience with clearer redirects and setup behavior

  **Improvements**

  * **Improved** playbook realtime updates by enabling full replica identity on the playbook tables
  * **Enhanced** dashboard navigation with name sorting, stable run links, and sidebar Recents syncing
  * **Refined** billing, usage history, and integration hover popups for clearer account details

  <span>Bug Fixes</span>

  * **Fixed** newly generated playbooks not showing up immediately
  * **Fixed** document realtime bootstrap and playbook document attachment loading
  * **Fixed** dashboard rows that looked not-ready even when playbooks had run history
  * **Fixed** stale login reload loops and stale session query parameters in sidebar navigation
</Update>

<Update label="May 2026" description="5.12.2026 | v0.2.8b">
  ## CLI, Mobile Chat, and Capability Index

  **New Features**

  * **Launched** the first Aident CLI package and unified CLI API endpoints
  * **Added** a mobile chat drawer so run details and chat context are easier to inspect on phones
  * **Introduced** Web3 login with Ethereum and Solana support through Supabase
  * **Added** failed-session status workflows to keep failure reports moving automatically

  **Improvements**

  * **Improved** capability indexing with event scaffolding, read cutover gates, and validation scripts
  * **Enhanced** mobile playbook surfaces with better tab switching, recents scrolling, run trace access, and drawer behavior
  * **Optimized** usage filters and pagination so longer histories are easier to navigate

  <span>Bug Fixes</span>

  * **Fixed** Cossistant translation timeline parts rendering incorrectly
  * **Fixed** running playbook islands losing tap targets
  * **Fixed** playbook deletion creating unwanted history
  * **Fixed** authenticated users being sent to the wrong place after login
</Update>

<Update label="April 2026" description="4.29.2026 | v0.2.6a">
  ## Usage Redesign & Billing Accuracy

  **New Features**

  * **Launched** a redesigned Usage page with clearer overview cards, history tables, date filters, and shared usage helpers
  * **Added** verification scripts for Composio billing overrides and capability VM billing behavior

  **Improvements**

  * **Improved** billing accuracy for `capabilities_execute` calls and first-party integration rates
  * **Enhanced** media generation billing so Seedance videos and Nano Banana images are tracked reliably
  * **Reduced** noisy Composio unavailable alerts so real connection problems are easier to spot

  <span>Bug Fixes</span>

  * **Fixed** the desktop playbook profile entry showing where it no longer belonged
  * **Fixed** usage page tests and credit balance command coverage around the redesigned billing views
</Update>

<Update label="April 2026" description="4.28.2026 | v0.2.6">
  ## Mobile Playbooks & Design System

  **New Features**

  * **Launched** mobile playbook, runs, chat, template, integrations, and profile surfaces
  * **Added** design system galleries for foundations, buttons, status navigation, logos, and website chrome
  * **Introduced** a maintenance banner that can be gated and shown globally
  * **Added** dashboard status select-all controls and clearer skills execution loading copy

  **Improvements**

  * **Improved** trigger editing with always-editable rows and clearer save/cancel behavior
  * **Enhanced** mobile composer, recents, and playbook spacing so core editor controls stay reachable
  * **Refined** virtual Composio execution and schema validation for cleaner integration runs

  <span>Bug Fixes</span>

  * **Fixed** hourly triggers forgetting their selected start hour
  * **Fixed** duplicate scheduler trigger executions
  * **Fixed** Composio action names being rewritten unexpectedly
  * **Fixed** several dashboard and playbook layout issues around progress panels, drawers, and pagination
</Update>

<Update label="April 2026" description="4.24.2026 | v0.2.5a">
  ## Event Triggers & Interactive Cards

  **New Features**

  * **Added** event trigger configuration directly in the playbook editor
  * **Introduced** interactive display cards for question flows, option lists, and confirmation prompts
  * **Shipped** a persistent navigation sidebar across app routes
  * **Launched** public MCP connections so external AI tools can connect more easily

  **Improvements**

  * **Improved** playbook run result cards, trigger session saves, and Run button feedback
  * **Enhanced** dashboard filters and failed-execution drawers so state is easier to understand after refreshes
  * **Refined** chat controls, image affordances, and playbook footer behavior around active runs

  <span>Bug Fixes</span>

  * **Fixed** trigger errors not opening the right integrations drawer
  * **Fixed** trigger save errors and hidden event default inputs
  * **Fixed** mobile login gating on the login page
  * **Fixed** awaiting-input sessions losing state for display tools
</Update>

<Update label="April 2026" description="4.23.2026 | v0.2.5">
  ## Aiden Talks Back & One-Click Connections

  **New Features**

  * **Launched** one-click connections so you can plug any external tool into your playbook without copy-pasting URLs or tokens
  * **Added** interactive cards in chat — Aiden can ask you a question, show options to pick from, or ask for a yes/no without pausing the run
  * **Introduced** event trigger setup directly in the playbook editor, so you don't need to leave the page to wire one up
  * **Shipped** a sidebar that stays put as you move between pages, so your recent playbooks are always one click away
  * **Built** cleaner result cards for runs — outputs are now easy to read at a glance instead of buried in chat

  **Improvements**

  * **Improved** the dashboard so runs that hung or got cancelled now show clearly as errors instead of silently sitting there
  * **Added** an instant indicator on the Run button so you know your click registered
  * **Enhanced** what happens when you run out of credits — runs that can't go through now say so plainly
  * **Optimized** how Aident notices when an app connection has expired so you can reconnect right away

  <span>Bug Fixes</span>

  * **Fixed** triggered runs occasionally not saving their progress
  * **Fixed** the "scroll to bottom" button overlapping the run status badge
  * **Fixed** the wrong icon showing on the dashboard's Running filter
  * **Fixed** duplicated playbooks accidentally sharing edits with the original
  * **Fixed** opening someone else's playbook not showing the "no access" message
</Update>

<Update label="April 2026" description="4.21.2026 | v0.2.4">
  ## New Playbook Page

  **New Features**

  * **Upgraded** Aiden to GPT-5.4 with extra-high reasoning, for sharper drafts and better problem solving
  * **Launched** a new home for every playbook — the canvas, the side chat, your triggers, and every past run all live on one page
  * **Added** a Runs view so you can scroll back through every past run and see exactly what Aiden did
  * **Shipped** a floating run indicator that shows a playbook running in the background without taking over the chat
  * **Introduced** quick actions on dashboard rows to rename, duplicate, and delete a playbook in two clicks
  * **Added** smarter integration suggestions that learn from how you use Aident and surface the apps you reach for most

  **Improvements**

  * **Improved** the sidebar to keep your recent playbooks visible with live running and unread indicators
  * **Enhanced** dashboard filters so they stay set when you refresh the page
  * **Added** a Contact Us link in the profile menu for faster help
  * **Optimized** how quickly the template gallery opens

  <span>Bug Fixes</span>

  * **Fixed** picking up a paused playbook not resuming where it left off
  * **Fixed** a new trigger not turning on automatically when you create it
  * **Fixed** images in chat disappearing after a page refresh
  * **Fixed** an occasional sign-in glitch right after returning from login
  * **Fixed** files not loading inside a past run's detail view
</Update>

<Update label="April 2026" description="4.3.2026 | v0.2.3">
  ## Playbooks Stand on Their Own

  **New Features**

  * **Introduced** playbooks as their own thing — your work is now saved with the playbook itself, not tied to a single chat
  * **Shipped** running a playbook from inside the editor — kick it off and watch the results without changing pages
  * **Added** email notifications when a triggered run finishes, with toggles so you can choose what's worth your inbox
  * **Launched** dedicated Dashboard, Capabilities, and Connections pages so navigating between your work, apps, and external access is faster
  * **Built** the ability to set up an event trigger before you've connected the app — finish the auth later
  * **Added** Amazon and Shopline integrations

  **Improvements**

  * **Improved** credit handling — credits are automatically refunded when a triggered run fails
  * **Enhanced** how Aiden finds the right integration action, with results that match what you actually have connected
  * **Added** a "Make a Copy" option so you can fork a playbook straight from the editor

  <span>Bug Fixes</span>

  * **Fixed** chat occasionally dropping messages mid-stream
  * **Fixed** edits to a playbook accidentally counting against your run credits
  * **Fixed** the integrations panel showing an app as connected when it wasn't
  * **Fixed** scheduled runs forgetting context they should have remembered
</Update>

<Update label="March 2026" description="3.18.2026 | v0.2.2">
  ## A New Dashboard & Run Emails

  **New Features**

  * **Launched** a redesigned Dashboard with sortable columns, filters, and clear status pills, so finding the right playbook is fast
  * **Added** email summaries when a triggered run finishes — know what happened without checking the app
  * **Shipped** the Connections page with tabs for Skills and MCP, alongside the integration catalog in Capabilities
  * **Introduced** "Make a Copy" for any playbook plus a button to jump back to the playbook from a run page
  * **Added** hover tooltips on dashboard status labels so you know exactly what each one means

  **Improvements**

  * **Improved** the path to setting up a trigger right after you finish drafting a playbook
  * **Enhanced** Aiden's reply speed for short questions
  * **Optimized** chat scrolling and animations for a smoother feel

  <span>Bug Fixes</span>

  * **Fixed** triggers not turning on after you save changes to them
  * **Fixed** the dashboard sometimes showing stale status for a running playbook
  * **Fixed** integrations not refreshing to "Connected" right after you signed in
</Update>

<Update label="March 2026" description="3.11.2026 | v0.2.1">
  ## A Smarter Aiden

  **New Features**

  * **Upgraded** Aiden to GPT-5 with automatic fallback to keep replies fast and reliable
  * **Launched** smarter search for actions — Aiden finds the right one even when you describe it loosely
  * **Added** transparent step-by-step thinking so you can see how Aiden gets to its answer
  * **Built** automatic context compaction in long chats, so older context stays useful without slowing things down
  * **Shipped** a "Mine" filter on integrations to jump straight to the apps you've already connected

  **Improvements**

  * **Improved** the hover cards on skills in chat to show what each one does and which app it uses
  * **Enhanced** input and output editing in the playbook editor with inline field controls
  * **Added** quick-look popups on integration skills throughout the editor

  <span>Bug Fixes</span>

  * **Fixed** newly-connected integrations not appearing in the picker right away
  * **Fixed** playbook generation occasionally failing
  * **Fixed** the integrations panel hanging on a loading spinner when you had nothing connected yet
</Update>

<Update label="February 2026" description="2.22.2026 | v0.2.0">
  ## Connections Hub & Credit Limits

  **New Features**

  * **Launched** the Connections page for chat bots and MCP tools, with app connections managed from the integration catalog
  * **Added** clearer credit limits — when you're out of credits, Aident tells you up front instead of failing a run halfway
  * **Shipped** a clearer **Test Run** flow across the editor
  * **Introduced** sort controls on the Dashboard so you can order playbooks by what matters to you
  * **Added** the option to pre-fill what an event trigger sends in, so triggered runs start with the right data

  **Improvements**

  * **Enhanced** the connect experience for integrations so signing in is a step or two shorter
  * **Improved** the dashboard drawer with quick dismiss for failed runs and pending approvals
  * **Added** search and pagination to the template gallery on Home

  <span>Bug Fixes</span>

  * **Fixed** the editor not fully reverting after a discard
  * **Fixed** triggers staying active after you deleted them
  * **Fixed** an issue that caused integration calls to fail with certain models
</Update>

<Update label="February 2026" description="2.12.2026 | v0.1.8">
  ## MCP Support & New Connections Page

  **New Features**

  * **Launched** MCP support so you can plug Aident into any MCP-compatible AI tool, like Claude or Cursor
  * **Added** Discord and Telegram chat bots — talk to Aiden right from your team's chat
  * **Shipped** extended MCP tools so an outside AI can browse, run, and manage your playbooks
  * **Introduced** quick-look popups for integrations and skills so you can see what they do without leaving the page
  * **Built** a Channels tab on the Connections page for bot connection setup

  **Improvements**

  * **Improved** the schedule trigger setup with a clearer flow and better default values
  * **Enhanced** the Dashboard to correctly show event trigger names and the time of each run

  <span>Bug Fixes</span>

  * **Fixed** deleted sessions disappearing for good instead of being recoverable
  * **Fixed** integrations showing under the wrong category
  * **Fixed** trigger default values not appearing in the quick-run prompt
</Update>

<Update label="February 2026" description="2.07.2026 | v0.1.7">
  ## Express Mode Launch

  **New Features**

  * **Launched** Express Mode - create simple workflows in seconds with a single AI generation step instead of multi-phase team building
  * **Introduced** lightweight playbook format with natural markdown structure - write workflows as numbered steps with inline skill references
  * **Built** dedicated Express playbook editor with redesigned UI for editing objectives, inputs, steps, and outputs directly
  * **Added** visual schema drawers for configuring input and output formats without touching markdown
  * **Shipped** scheduler trigger support for Express playbooks with timezone-aware cron scheduling
  * **Implemented** automatic skill discovery during generation using parallel capability searches across integrations

  **Improvements**

  * **Added** smart validation that requires at least one resolvable skill reference while allowing pure reasoning steps
  * **Implemented** automatic toolset derivation from inline skill tags to configure agent permissions
  * **Fixed** trigger service compatibility to properly read and execute Express playbooks
  * **Improved** authentication status visibility in Express playbook rendering
  * **Enhanced** Composio trigger integration for automated workflow execution
  * **Optimized** message delivery in Express execution agent for clearer final results

  **What is Express Mode?**

  Express Mode is a fast-path workflow creation system perfect for simple, linear workflows:

  * **1 LLM call** generates your complete workflow (vs. 6-phase multi-agent generation)
  * **No compilation** needed - workflows are ready to run immediately
  * **Single agent execution** - faster runtime without coordination overhead
  * **Best for**: Linear workflows under \~5 steps, single-domain tasks, quick prototypes

  Traditional Full Mode remains available for complex multi-agent workflows requiring parallel work, QA gates, and cross-domain coordination.
</Update>

<Update label="February 2026" description="2.03.2026 | v0.1.6">
  ## Playbook Dashboard & Triggers

  **New Features**

  * **Launched** playbook dashboard with info drawer, trigger management, and execution history
  * **Shipped** integration webhook integration for external trigger events, such as trigger your playbook run via receiving an email!
  * **Added** Shipment Tracking from Gmail to Google Calendar template

  **Improvements**

  * **Improved** playbook trigger handling to auto-deactivate incompatible triggers after schema changes

  <span>Bug Fixes</span>

  * **Fixed** flow chart not updating or regenerating for ready state playbooks
  * **Fixed** workflow auto-compilation when changes don't require recompilation
</Update>

<Update label="February 2026" description="2.01.2026 | v0.1.5">
  ## Instagram Integration & System Improvements

  **New Features**

  * **Launched** Instagram integration with toolset and tools
  * **Added** credit system with balance display and management
  * **Added** playbook team shared document tools for Project Manager agent

  **Improvements**

  * **Enhanced** integrations modal with dynamic loading, category filtering, and better search
  * **Enhanced** loop detection logic with stricter termination rules to prevent infinite execution
  * **Improved** error messages provided to AI models for better debugging

  <span>Bug Fixes</span>

  * **Fixed** Composio toolsets authentication config requirements and validation
  * **Fixed** playbook generation by switching to Gemini 3 Pro model
  * **Fixed** session status handling in chat service initialization
</Update>

<Update label="January 2026" description="1.30.2026 | v0.1.4">
  ## Platform Stability & Performance

  **Improvements**

  * **Enhanced** overall platform stability and performance
  * **Improved** system reliability across core features

  <span>Bug Fixes</span>

  * **Fixed** various bugs and issues reported by users
  * **Improved** error handling and recovery
</Update>

<Update label="January 2026" description="1.22.2026 | v0.1.3">
  ## New Home UI & Enhanced Chat

  **New Features**

  * **Built** Completely New Home UI with updated sidebar, session management, and asset upload
  * **Built** image upload and preview in chat composer with expanded view

  **Improvements**

  * **Improved** session sorting by updated time

  <span>Bug Fixes</span>

  * **Fixed** messages not loading in chat threads
  * **Fixed** image preview display in chat thread
  * **Fixed** user profile clickability issue
  * **Fixed** login page back button to redirect to aident.ai instead of browser history
</Update>

<Update label="January 2026" description="1.15.2026 | v0.1.2">
  ## Monitoring & Better UX

  **Improvements**

  * **Reduced** UX confusion we heard from early users
  * **Added** better monitoring for bugs & issues

  <span>Bug Fixes</span>

  * **Fixed** bugs for connecting Github & Lark
  * **Fixed** agent sessions stuck in bad state
  * **Fixed** authentication in agent chat thread
</Update>

<Update label="December 2025" description="12.30.2025 | v0.1.1">
  ## Stability & Usability

  **Improvements**

  * **Improved** UI in multiple places
  * **Improved** infrastructure for faster & better product release
  * *Stabilized* agent infrastructure
</Update>

<Update label="December 2025" description="12.22.2025 | v0.1.0">
  ## Skills & New Chat Experience

  **New Features**

  * **Launched Skills** - reusable building blocks that make your agents smarter and more capable
  * **New Playbook Editor** - rich text editing with better formatting and a cleaner writing experience
  * **Upgraded to Gemini Flash 3.0** - faster, smarter responses across all agents
  * **Chat History Drawer** - easily review past conversations with collapsible sections
  * **In-Chat Integrations** - connect your apps directly from the conversation without leaving the chat
  * **Auto-Retry** - automatically resend messages after connecting missing integrations

  **Improvements**

  * **Cleaner chat layout** with better message formatting and image support
  * **Smoother test runs** with live status updates and resizable panels
  * **New loading animations** throughout the platform

  <span>Bug Fixes</span>

  * **Fixed** toolsets not removing properly
  * **Fixed** sessions stopping unexpectedly
  * **Fixed** certain templates not loading correctly on first open
  * **Improved** connection stability for longer-running workflows
</Update>

<Update label="November 2025" description="11.29.2025 | v0.0.10">
  ## In-App Support & UI Polish

  **New Features**

  * **Launched** In-app feedback system with Interactive Support Chat powered by
    Cossistant
  * **Enhanced Flow Chart** visualization with movable charts and better Agent Roster display
    names
  * **Added** parallel tool call infrastructure for **Faster Multi-tool Execution**
  * **Improved** template detail modal with zoomable and pannable flow charts

  <span>Bug Fixes</span>

  * **Fixed** Trigger **Creation and Update** handling in playbook editor
  * **Resolved** flow chart rendering issues with proper box lines and template display
  * **Fixed** some templates may not render succesfully
  * **Enhanced** Agent Roster badge hover positioning and sizing
  * **Fixed** issues with Google Sheets to Sending Emails template execution
  * **Improved Editor Responsiveness** with proper gap filling when closing panels
  * **Improved** UI for confirmation modal backdrop rendering on home page
</Update>

<Update label="November 2025" description="11.22.2025 | v0.0.9">
  ## Session Continuity & Gemini 3.0 Pro

  **New Features**

  * **Upgraded** to Gemini 3.0 Pro! - Aiden is now Faster and more capable reasoning
  * **Implemented** robust session continuity with exponential backoff retry logic
  * **Added** timezone support across all APIs and agent configurations
  * **Enabled Test Run** session caching in localStorage for better UX
  * **Shipped** GMail Event and Deadline Extractions to Google Calendar automation template

  <span>Bug Fixes</span>

  * **Fixed** unexpected session stops with improved **Resume Handling**
  * **Resolved** dead locks in playbook skill service with Redis locking
  * **Improved** internal Agent Roster message handling to reduce noise
  * **Enhanced** Zod error wrapping for better error recovery
  * **Fixed** image upload issues with our **Twitter Posts Automation** template issues
</Update>

<Update label="November 2025" description="11.16.2025 | v0.0.8">
  ## Execution Plans & Enhanced Editor

  **New Features**

  * **Launched** **Execution Plan** - configure input schemas, output formats, and triggers
    visually in the playbook editor now
  * **Added Scheduler Trigger** support with timezone-aware scheduling for automated
    workflows
  * **Built LlamaParser Integration** for document parsing and extraction
  * **Shipped Test Run Panel** with resizable interface and auto-message injection
  * **Added Version Link** in user profile dropdown for quick changelog access
  * **Implemented** bug report screenshot upload for better **issue reporting**
  * **Enhanced** capability group library with **Skill Action Search** and filtering capabilities
  * **Added** smart embedding reindexing with cache invalidation for **faster capability group semantic search**

  <span>Bug Fixes</span>

  * **Fixed** workflows auto-compiling when changes don't require recompilation
  * **Improved** capability group disconnection detection and reconnection flows
  * **Fixed** template background images, icons, and category rendering in Home Dashboard
  * **Enhanced** query parameter state sync across playbook editor components
  * **Fixed** asset ID handling for external service uploads
  * **Improved error recovery** for invalid capability group handling
</Update>

<Update label="November 2025" description="11.09.2025 | v0.0.7">
  ## Capability group Library & Semantic Search

  **New Features**

  * **Shipped** semantic search for skills and tools - Aiden now finds the right one by understanding what you need
  * **Launched** comprehensive docs site with tutorials and guides
  * **Added** beautiful custom icons throughout the platform
  * **Enabled** 15+ new integrations: Notion, Slack, GitHub, Google Sheets, LinkedIn, Calendly, Dropbox, Airtable, Box, Lark
  * **Improved** tool discovery with intelligent search suggestions

  <span>Bug Fixes</span>

  * **Fixed** privacy issue where sessions could be visible to other users under rare conditions
  * **Improved** tool naming for better AI Agents compatibility
  * **Enhanced** integration reliability with better error recovery
  * **Fixed** Twitter login and connection issues
</Update>

<Update label="October 2025" description="10.21.2025 | v0.0.6">
  ## Agent Team Runtime & Template System

  **New Features**

  * **Added** many quick-start templates to get your workflows running instantly
  * **Built** playbook nodes visualization - see how your Roster team will execute tasks in a flow chart
  * **Integrated** AI-powered image generation with Google Gemini
  * **Enabled** Twitter posting with media uploads
  * **Improved** asset management for easier file handling

  <span>Bug Fixes</span>

  * **Fixed** workflows auto-compiling when they shouldn't in some cases
  * **Improved** error messages to be more helpful
  * **Enhanced** session reliability and performance
  * **Fixed** image handling issues
  * **Corrected** tool response processing
</Update>

<Update label="September 2025" description="09.24.2025 | v0.0.5">
  ## Multi-Agent Collaboration & Integrations

  **New Features**

  * **Built** a multi-agent Roster team that will help you automate together on complex tasks
  * **Enabled** @mentions and internal docs so agents can collaborate naturally
  * **Launched** one-click deployment for your workflows

  <span>Bug Fixes</span>

  * **Fixed** tool connection issues
  * **Improved** error handling throughout the platform
  * **Enhanced** coordination between multiple agents
  * **Fixed** workflow state saving
  * **Corrected** team member management
</Update>

<Update label="August 2025" description="08.23.2025 | v0.0.3">
  ## Playbook Compilation & AI SDK Upgrade

  **New Features**

  * **Launched** auto-compilation system for your workflows
  * **Added** live deployment progress tracking with milestones
  * **Improved** chat interface with better message rendering
  * **Added** thinking visualization to see agent reasoning

  <span>Bug Fixes</span>

  * **Fixed** compilation auto-triggering issues when refreshes entire page
  * **Improved** session status handling and polling
  * **Enhanced** message parsing and streaming
  * **Fixed** UI crashes and some rendering issues
  * **Corrected** integration connection flows
</Update>

<Update label="July 2025" description="07.28.2025 | v0.0.1">
  ## The Beginning

  * **Shipped** our first MVP version, with basic editor and chat features
  * **Added** Google OAuth login working with Supabase Auth
  * **Integrated** LLM streaming responses, now messages flow in real-time looks satisfying!
  * **Integrated** 5 initial integrations (GMail, GCal, Github, GDocs, GDrive)
</Update>


# How It Works
Source: https://docs.aident.ai/documentation/how-it-works

From a plain-English description to a running automation

Aident turns a short description into a working automation. Here's what happens from the moment you land on **Home** to the point where your playbook runs on its own.

## The Journey

<Steps>
  <Step title="Describe your job on Home">
    The **Home** screen in the left sidebar has one big input. Type what you want done in plain English — as specific or as general as you like. When you send it, Aident opens the editor for a new playbook.

    <Frame>
      <img alt="The Home screen with the big input where you describe a new automation" />
    </Frame>
  </Step>

  <Step title="Aiden drafts the playbook">
    **Aiden** is the chat in the side panel of the editor. It writes the first draft of your playbook — the **🎯 Goal**, the **⚙️ Integrations** it needs, and the **📄 Plan** (an **Input** card, ordered **Steps**, and an **Output** card). You can edit any section directly, or ask Aiden to rework it.

    <Frame>
      <img alt="Aiden drafting a playbook in the editor with Goal, Integrations, and Plan sections" />
    </Frame>
  </Step>

  <Step title="Connect the apps you need">
    Look at the integration pills under **⚙️ Integrations**. Orange means you haven't signed in yet — click the pill to connect. Red means Aiden couldn't resolve a reference; click **Ask Agent to Fix** and Aiden will repair it. White means ready to go.

    <Frame>
      <img alt="Integration pills in the editor showing connection status for each app" />
    </Frame>
  </Step>

  <Step title="Set up a Trigger — or just run it now">
    Open the **Triggers** tab at the top to choose when your playbook runs. Pick a **Time Trigger** for a schedule, or an **External Event** trigger to run when something happens in Gmail, Slack, or another connected app. You can also run it on demand with **Test Run** while you're testing.

    <Frame>
      <img alt="Triggers tab showing Time Trigger and External Event options" />
    </Frame>
  </Step>

  <Step title="Watch it on the Dashboard">
    The **Dashboard** in the left sidebar shows every playbook you've built, its current status, active trigger state, and last run. Open a playbook to review its full results history.

    <Frame>
      <img alt="Dashboard listing your playbooks with status labels and recent runs" />
    </Frame>
  </Step>

  <Step title="Keep refining with Aiden">
    Aiden stays available in the side chat of the editor. Ask it to add a step, tighten the Goal, switch integrations, or explain why something ran the way it did. The playbook updates in place.

    <Frame>
      <img alt="Aiden refining a playbook in the side chat of the editor" />
    </Frame>
  </Step>
</Steps>

## The Three Tabs

Every playbook has three tabs at the top of its page. You'll switch between them as you design, schedule, and review.

<div>
  <a href="/documentation/playbooks">
    <div>
      <img alt="" />

      <span>Playbook</span>
    </div>

    <p>
      The main canvas, plus Aiden in the side chat. Design and refine here.
    </p>

    <ul>
      <li>🎯 Goal</li>
      <li>⚙️ Integrations</li>
      <li>📄 Plan</li>
    </ul>
  </a>

  <a href="/documentation/triggers">
    <div>
      <img alt="" />

      <span>Triggers</span>
    </div>

    <p>
      Configure when the playbook runs — a Time Trigger for a schedule, or an External Event for something in a connected app. The Input and Output live here too.
    </p>
  </a>

  <div>
    <div>
      <img alt="" />

      <span>Results</span>
    </div>

    <p>
      The full history of past runs. Click any run to see what Aiden did, the input it received, and the output it produced.
    </p>
  </div>
</div>

## Dashboard Status at a Glance

When you look at your list on the Dashboard, each playbook shows one of these labels.

<AccordionGroup>
  <Accordion title="Healthy — triggers active and running" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-healthy.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=47446e10038380b1ca7b1c57aca439f5">
    The playbook has active triggers and recent runs are succeeding. Nothing to do.
  </Accordion>

  <Accordion title="Running — execution in progress" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-running.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=f9cab42bce34182917c7388345d539db">
    A run is happening right now. Open it to watch live.
  </Accordion>

  <Accordion title="Awaiting — needs your input or approval" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-awaiting.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=ce9fa24a37e36089866891bb4c538b3e">
    The playbook paused and is waiting on you. Open it to respond.
  </Accordion>

  <Accordion title="Failed — last execution errored" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-failed.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=3b71ab053eb3f71ec9d08c1981fa4f53">
    The last run didn't finish. Open it to see what went wrong, or ask Aiden to fix it.
  </Accordion>

  <Accordion title="Not Ready — no active triggers configured" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-not-ready.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=f6f40050db0c9b6078f10724c82ba2f9">
    The playbook exists but has no trigger turned on. Open the Triggers tab to add one.
  </Accordion>
</AccordionGroup>

## Next Steps

<CardGroup>
  <Card title="Get Started" icon="rocket" href="/get-started">
    Build your first playbook from Home
  </Card>

  <Card title="Playbooks" icon="book" href="/documentation/playbooks">
    Tour the three sections of the canvas
  </Card>

  <Card title="Triggers" icon="clock" href="/documentation/triggers">
    Choose when your playbook runs
  </Card>

  <Card title="Connect Your Tools" icon="plug" href="/documentation/integrations">
    Sign in to the apps your playbook uses
  </Card>
</CardGroup>


# Install Aident Skill
Source: https://docs.aident.ai/documentation/install-aident-skill

Add Aident to your AI assistant in one command

Install the single Aident skill to teach your AI assistant how to use Aident Loadout integrations.

## One-Command Install

If your agent supports [Skills.sh](https://skills.sh), install with a single command:

```bash theme={null}
npx skills add aident-ai/aident-skill
```

This installs local agent guidance only. It does not automatically configure MCP. For first-time setup, ask your agent:

```text theme={null}
Follow https://aident.ai/SETUP.md
```

## Manual Setup

If you want MCP instead of the CLI, add the Aident Loadout server URL to your client's MCP configuration file.

<AccordionGroup>
  <Accordion title="Claude Code">
    Run in your terminal:

    ```bash theme={null}
    claude mcp add --transport http aident https://loadout.aident.ai/mcp
    ```

    Or add to `~/.claude.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "type": "http",
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Claude Desktop">
    Add to your config file:

    * **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
    * **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "command": "npx",
          "args": ["-y", "mcp-remote", "https://loadout.aident.ai/mcp"]
        }
      }
    }
    ```

    Or on **Pro/Max/Team/Enterprise** plans: **Settings → Connectors → Add** and enter `https://loadout.aident.ai/mcp`.
  </Accordion>

  <Accordion title="Cursor IDE">
    Add to `.cursor/mcp.json` in your project root:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="VS Code (Copilot)">
    Add to `.vscode/mcp.json` in your project root:

    ```json theme={null}
    {
      "servers": {
        "aident": {
          "type": "http",
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Windsurf">
    Add to `~/.codeium/windsurf/mcp_config.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "serverUrl": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Codex">
    Run in your terminal:

    ```bash theme={null}
    codex mcp add aident --url https://loadout.aident.ai/mcp
    ```

    Or add to `~/.codex/config.toml` (or `.codex/config.toml` for a trusted project):

    ```toml theme={null}
    [mcp_servers.aident]
    url = "https://loadout.aident.ai/mcp"
    ```

    Verify the server is configured:

    ```bash theme={null}
    codex mcp list
    ```
  </Accordion>

  <Accordion title="Gemini CLI">
    Add to `~/.gemini/settings.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "httpUrl": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Other Clients">
    Any MCP-compatible client (Goose, Kiro, OpenCode, Antigravity, Factory, etc.) can connect using the server URL `https://loadout.aident.ai/mcp`. Refer to your client's documentation for where to add MCP server configurations.
  </Accordion>
</AccordionGroup>

## Authenticate

On first use, your MCP client will automatically:

1. Discover Aident's OAuth endpoints
2. Open your browser for login (Google OAuth or magic link)
3. Store your access token locally

No manual token management needed — the client handles everything.

## Verify

Ask your AI assistant to use Aident Loadout. Requests should route to external integrations, Vault status, execution, and audit history.

<Tip>
  Try asking: "Use Aident Loadout to check my integration status"
</Tip>

## Next Steps

* [MCP Server Setup](/documentation/mcp-server-setup) — Detailed setup guide with troubleshooting
* [MCP API Reference](/documentation/mcp-api-reference) — Full reference for the Aident tool suite


# Integrations
Source: https://docs.aident.ai/documentation/integrations

The apps your playbook uses, and how to connect them

An **integration** is an app your playbook talks to — Gmail to send email, Slack to post a message, Google Sheets to read a row, and so on. Aident keeps the connection to each app so your playbook can act on your behalf.

You'll run into integrations in two places: right inside the **Playbook editor**, and on the **Capabilities** page in the left sidebar.

***

## In the Playbook Editor

On the editor canvas, the **⚙️ Integrations** section shows one pill for every app the playbook uses. The pill's color tells you what to do.

<AccordionGroup>
  <Accordion title="White pill — ready to use">
    Connected and working. The playbook can use the app as-is.
  </Accordion>

  <Accordion title="Orange pill — needs connecting">
    You haven't signed in to this app yet. Click the pill, finish the sign-in, and come back.
  </Accordion>

  <Accordion title="Red pill — unresolved">
    Something about this reference can't be matched to a real app or skill. Click **Ask Agent to Fix** and Aiden will repair it.
  </Accordion>
</AccordionGroup>

You don't add integrations by hand — Aiden picks them based on your Goal and your Steps. Change the plan and the pills update themselves.

***

## Connecting an App

The first time a playbook needs an app, you'll be asked to sign in.

<Steps>
  <Step title="Click the integration pill">
    An orange pill in the editor, or the **Connect** button on the Capabilities page, will open a standard sign-in window.
  </Step>

  <Step title="Sign in with your account">
    Use the account you want the playbook to act from. This is a normal OAuth flow from the provider.
  </Step>

  <Step title="Approve access">
    Grant the permissions the screen asks for. Aident asks for what your current playbook or future actions may need.
  </Step>

  <Step title="You're done">
    The pill turns white and the playbook can use the app right away.
  </Step>
</Steps>

<Info>
  Aident never sees your password. Sign-in goes straight to the provider; Aident only receives a token you can revoke at any time.
</Info>

***

## The Capabilities Page

Open **Capabilities** from the left sidebar. The header reads "Capabilities" and shows the integrations catalog. Most connection work happens on the **Integrations** tab; depending on your account role, you may also see admin-oriented tabs.

<Tabs>
  <Tab title="Integrations">
    Browse apps Aident can connect to, search the catalog, and use **Connect** or **Disconnect** on each integration.
  </Tab>

  <Tab title="Vault">
    Admin users can manage API-key style credentials for integrations that use stored secrets instead of an OAuth sign-in.
  </Tab>

  <Tab title="Audit">
    Review recent capability usage.
  </Tab>
</Tabs>

The **Connections** page is separate. It manages MCP access, reusable skills, channels, and sandbox-related setup.

***

## Managing Connections

<AccordionGroup>
  <Accordion title="Reconnect an app">
    If a connection expires, the pill turns orange in the editor. Click it to sign in again and you're back in business.
  </Accordion>

  <Accordion title="Disconnect an app">
    Go to **Capabilities** and click **Disconnect** on the integration you want to remove. Any playbook that uses it will need to be reconnected before it runs again.
  </Accordion>

  <Accordion title="Resync a channel">
    Open **Connections** and use the channel controls if a bot's state looks out of date.
  </Accordion>
</AccordionGroup>

<Warning>
  Disconnecting affects every playbook that relies on the app. Make sure no scheduled runs depend on it before you revoke access.
</Warning>

***

## What Aident Supports

Aident has a growing catalog across communication (Gmail, Slack, Microsoft Teams, Discord, Telegram), productivity (Google Drive, Docs, Sheets, Calendar, Notion, Lark, Box), developer tools (GitHub), AI and media (Fal AI, Gladia), search (Exa, SerpAPI, Firecrawl), social (X), and storage (Dropbox) — among others.

<Tip>
  Open **Capabilities** in the sidebar to see the full, up-to-date integration catalog. That's the source of truth, not any static table in the docs.
</Tip>

***

## Security

* Sign-in goes directly to the provider over OAuth — Aident never sees your password.
* Aident only asks for the permissions your playbook actually needs.
* You can disconnect any integration from the **Capabilities** page at any time.

***

## Next Steps

<CardGroup>
  <Card title="Playbooks" icon="book" href="/documentation/playbooks">
    See where integration pills live on the editor canvas
  </Card>

  <Card title="Skills" icon="bolt" href="/documentation/skills">
    Learn how steps reference actions that use your integrations
  </Card>
</CardGroup>


# Aident Loadout
Source: https://docs.aident.ai/documentation/loadouts

Equip AI agents with governed capabilities, delegated credentials, policies, and audit receipts

Aident Loadout gives every agent the right capabilities for the job. A loadout is a curated set of capabilities,
credentials, policies, and audit rules that an agent can carry into a task.

Open **Aident Loadout** at [loadout.aident.ai/home](https://loadout.aident.ai/home).

***

## What Loadout Includes

<CardGroup>
  <Card title="Integrations" icon="plug">
    Browse accessible integrations, connect OAuth accounts, and open an integration drawer to inspect its available
    actions.
  </Card>

  <Card title="Vault" icon="key">
    Admin users can add, rotate, or delete delegated credentials for integrations that do not use OAuth.
  </Card>

  <Card title="Audit" icon="clipboard-list">
    Review action-call history, status, source, and receipts for agent work.
  </Card>

  <Card title="Loadouts" icon="shield-check">
    Governed capability sets let agents use integrations without directly handling raw provider credentials.
  </Card>
</CardGroup>

***

## Set Up an Agent Harness

Use the same setup prompt in any supported coding agent or automation harness:

```text theme={null}
Setup https://aident.ai/SKILL.md
```

For this coding-agent flow, no separate CLI or MCP configuration is required. Paste the prompt into your agent session
and let the agent fetch the remote Aident skill instructions.

On first-time setup, the agent will ask whether to scan local MCP or agent integration configs for supported Loadout
migration candidates. The scan is opt-in, redacts secret-like values, and lets you connect all supported matches or only
selected integrations through Loadout Vault/OAuth.

Supported harnesses include:

* Codex
* Claude Code
* OpenClaw
* HermesAgent
* OpenCode
* Cursor
* Pi
* Cline
* Windsurf
* VS Code
* n8n
* MCP-compatible agent clients

***

## Connect Integrations

<Steps>
  <Step title="Open Integrations">
    Go to `/loadout/integrations` or select **Integrations** from the Loadout navigation.
  </Step>

  <Step title="Search the catalog">
    Search by integration name, capability, or provider. Use category filters to narrow the catalog.
  </Step>

  <Step title="Open integration details">
    Click an integration to open the drawer. The drawer shows metadata and the available actions for that integration.
  </Step>

  <Step title="Connect or disconnect">
    Use the drawer or card action to start the provider flow. Connected integrations show their ready state and can be
    disconnected when the provider supports it.
  </Step>
</Steps>

<Info>
  Built-in and Aident-managed integrations do not require a user-managed connection. They show as available without a
  connect button.
</Info>

***

## Manage Vault Credentials

The **Vault** tab is available to admin users. Use it for integrations backed by stored credentials rather than OAuth.

<Warning>
  Rotating or deleting a credential affects every action that depends on that integration. Check active agents and
  scheduled work before removing a credential.
</Warning>

***

## Review Audit History

Open **Audit** to review recent action calls. Use the date filters to inspect a time range, then review:

* Date and time
* Action name
* Integration account
* Source
* Status
* Credits used

Audit history is the fastest way to confirm which integration action ran and whether it completed successfully.

***

## Next Steps

<CardGroup>
  <Card title="Integrations" icon="plug" href="/documentation/integrations">
    Learn how Aident connects provider accounts
  </Card>

  <Card title="MCP Server Setup" icon="terminal" href="/documentation/mcp-server-setup">
    Connect Aident to MCP-compatible clients
  </Card>
</CardGroup>


# MCP API Reference
Source: https://docs.aident.ai/documentation/mcp-api-reference

Reference for the Aident MCP tool suite

The Aident MCP Server exposes the full suite of Aident tools across auth, capabilities, skills, integrations, playbooks, templates, and dashboard. All tools are accessible through any MCP-compatible client after [setup](/documentation/mcp-server-setup).

## Auth Tools

### auth\_status

Check your current authentication status. Always accessible, even without scopes.

**No input required.**

### auth\_logout

Revoke your access token and log out. Use this to switch accounts — reconnect after logout to sign in with a different account.

**No input required.**

## Capabilities, Skills & Discovery

### Capability execution contract

Design PRD: Actions & Skills Naming Alignment Phase 4.

`capabilities_execute` is the execution tool. It accepts `{ name, input }`; integration Actions must use `${integration_type}:${integration_name}:${action_name}` names such as `composio:gmail_tools:gmail_send_email`. Legacy dotted or typed-dot action refs are rejected with typed-format guidance. Plain authored Skill names route directly through Skill execution.

Integration Actions return the Action output directly without `durationMs` or `executionTimeMs`. Skill execution returns `{ success, output?, error?, executionTimeMs }`.

### capabilities\_search

Search capabilities by query, including Actions, Skills, and integrations. Supports keyword, semantic, and hybrid search modes.

| Parameter | Type      | Required    | Description                                                         |
| --------- | --------- | ----------- | ------------------------------------------------------------------- |
| `query`   | string    | Yes (min 2) | Search query — supports `*`, `-`, `"phrases"`, `\|` operators       |
| `types`   | string\[] | No          | Filter by type: `integration`, `skill`                              |
| `scope`   | object    | No          | Scope filtering: `{ integrationId?, category? }`                    |
| `mode`    | string    | No          | Search mode: `keyword`, `semantic`, or `hybrid` (default: `hybrid`) |
| `limit`   | number    | No          | Max results, 1-50 (default: 20)                                     |

### capabilities\_get

Look up details for a known integration or capability by exact name. By default only `description` is returned; request additional parts as needed.

MCP exposes this as `capabilities_get`; internally it delegates to the capability runtime's `capability_read_details` tool.

| Parameter | Type      | Required | Description                                                                                                                                                        |
| --------- | --------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `name`    | string    | Yes      | Exact integration ID (e.g., `composio:gmail_tools`) or capability name (e.g., `composio:gmail_tools:gmail_send_email`)                                             |
| `parts`   | string\[] | No       | Sections to include. Integration: `description`, `agentFriendlyInstructions`, `availableSkills`. Skill: `description`, `inputSchema`, `outputSchema`, `sourceCode` |

### capabilities\_execute

Execute a capability by name. Accepts either a skill name (e.g., `send_email_skill`) or a typed integration action in `${integration_type}:${integration_name}:${action_name}` form (e.g., `composio:gmail_tools:gmail_send_email`). If required integrations are missing, returns connection instructions.

| Parameter | Type   | Required | Description                                                   |
| --------- | ------ | -------- | ------------------------------------------------------------- |
| `name`    | string | Yes      | Capability name — skill name or typed integration action name |
| `input`   | object | No       | Input data matching the capability's schema                   |

Response shape:

* Skills return `{ success, output?, error?, executionTimeMs }`.
* Integration Actions return the Action output directly. This shape omits `durationMs` and `executionTimeMs`.

## Vault Tool

### vault

Check, connect, or disconnect third-party integrations through Aident Vault.

| Parameter       | Type                    | Required | Description                                                      |
| --------------- | ----------------------- | -------- | ---------------------------------------------------------------- |
| `action`        | string                  | Yes      | `status`, `connect`, or `disconnect`                             |
| `integrationId` | string                  | No       | Specific integration ID. Required for `connect` and `disconnect` |
| `confirmed`     | boolean                 | No       | Confirm credit usage for first-party integrations                |
| `credentials`   | Record\<string, string> | No       | User-provided credentials (API keys, tokens, etc.)               |

## Playbook Tools

### playbooks\_list

List your playbooks with status and trigger info.

| Parameter | Type   | Required | Description               |
| --------- | ------ | -------- | ------------------------- |
| `limit`   | number | No       | Max results (default: 20) |
| `offset`  | number | No       | Pagination offset         |

### playbooks\_get

Get playbook details including content and trigger configuration.

| Parameter    | Type   | Required | Description                |
| ------------ | ------ | -------- | -------------------------- |
| `playbookId` | string | Yes      | Playbook editor session ID |

### playbooks\_generate

Generate a new playbook from a natural language description.

| Parameter | Type   | Required | Description                                            |
| --------- | ------ | -------- | ------------------------------------------------------ |
| `prompt`  | string | Yes      | Description of the playbook to generate (min 10 chars) |

### playbooks\_execute

Execute a playbook. Returns an execution session ID for tracking.

| Parameter    | Type   | Required | Description                          |
| ------------ | ------ | -------- | ------------------------------------ |
| `playbookId` | string | Yes      | Playbook editor session ID           |
| `input`      | object | No       | Optional input data for the playbook |

### playbooks\_trigger\_activate

Enable or disable a playbook's trigger.

| Parameter    | Type    | Required | Description                              |
| ------------ | ------- | -------- | ---------------------------------------- |
| `playbookId` | string  | Yes      | Playbook editor session ID               |
| `enabled`    | boolean | Yes      | Whether to enable or disable the trigger |

### playbooks\_history

Get execution history for a specific playbook.

| Parameter    | Type   | Required | Description                |
| ------------ | ------ | -------- | -------------------------- |
| `playbookId` | string | Yes      | Playbook editor session ID |
| `limit`      | number | No       | Max results (default: 10)  |

## Template Tools

### templates\_search

Search for playbook templates by keyword or category.

| Parameter  | Type   | Required | Description        |
| ---------- | ------ | -------- | ------------------ |
| `query`    | string | No       | Search query       |
| `category` | string | No       | Filter by category |

### templates\_list

List available templates with optional category filtering.

| Parameter  | Type   | Required | Description               |
| ---------- | ------ | -------- | ------------------------- |
| `category` | string | No       | Filter by category        |
| `limit`    | number | No       | Max results (default: 20) |
| `offset`   | number | No       | Pagination offset         |

### templates\_get

Get detailed template information including description and required integrations.

| Parameter    | Type   | Required | Description |
| ------------ | ------ | -------- | ----------- |
| `templateId` | string | Yes      | Template ID |

### templates\_instantiate

Create a new playbook from a template.

| Parameter    | Type   | Required | Description                      |
| ------------ | ------ | -------- | -------------------------------- |
| `templateId` | string | Yes      | Template ID                      |
| `name`       | string | No       | Custom name for the new playbook |

## Dashboard Tools

### dashboard\_overview

List playbooks with active triggers or running executions.

**No input required.**

### executions\_get

Get execution details including status, messages, and results.

| Parameter     | Type   | Required | Description          |
| ------------- | ------ | -------- | -------------------- |
| `executionId` | string | Yes      | Execution session ID |

### executions\_list

List recent executions across all playbooks.

| Parameter | Type   | Required | Description               |
| --------- | ------ | -------- | ------------------------- |
| `limit`   | number | No       | Max results (default: 20) |
| `offset`  | number | No       | Pagination offset         |

### executions\_messages

Get simplified chat messages for a running or completed execution. Useful for polling progress.

| Parameter   | Type   | Required | Description                    |
| ----------- | ------ | -------- | ------------------------------ |
| `sessionId` | string | Yes      | Session ID to get messages for |

## Scopes

Tools are organized into permission scopes. Access is granted during OAuth authorization:

| Scope                             | Tools                                                                        |
| --------------------------------- | ---------------------------------------------------------------------------- |
| `capabilities:search`             | capabilities\_search                                                         |
| `capabilities:read`               | capabilities\_get                                                            |
| `capabilities:execute`            | capabilities\_execute                                                        |
| `integrations:read`               | vault status                                                                 |
| `integrations:connect`            | vault connect, vault disconnect                                              |
| `playbooks:read`                  | playbooks\_list, playbooks\_get, playbooks\_history                          |
| `playbooks:create`                | playbooks\_generate                                                          |
| `playbooks:execute`               | playbooks\_execute                                                           |
| `playbooks:manage`                | playbooks\_trigger\_activate                                                 |
| `templates:read`                  | templates\_search, templates\_list, templates\_get                           |
| `templates:use`                   | templates\_instantiate                                                       |
| `dashboard:read`                  | dashboard\_overview, executions\_get, executions\_list, executions\_messages |
| `sandbox:read`, `sandbox:execute` | Used by sandbox-capable tools                                                |

The `auth` tool is always accessible regardless of scopes.


# MCP Server Setup
Source: https://docs.aident.ai/documentation/mcp-server-setup

Connect Claude Code, Cursor, or any MCP-compatible tool to Aident

Connect your favorite AI assistant to Aident Loadout integrations and Playbook workflows through the Model Context Protocol (MCP).

## Prerequisites

* Active [Aident account](https://app.aident.ai)
* Connected integrations in [Aident Loadout](https://loadout.aident.ai/integrations)
* An MCP-compatible client

## Step 1: Add Server URL

Use the product MCP URL for the package you need. Loadout is the default integration surface:

```
https://loadout.aident.ai/mcp
```

For Playbook workflows, configure a separate MCP server at:

```
https://app.aident.ai/playbook/mcp
```

<AccordionGroup>
  <Accordion title="Claude Code">
    Run in your terminal:

    ```bash theme={null}
    claude mcp add --transport http aident https://loadout.aident.ai/mcp
    ```

    Or add to `~/.claude.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "type": "http",
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Claude Desktop">
    Add to your config file:

    * **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
    * **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "command": "npx",
          "args": ["-y", "mcp-remote", "https://loadout.aident.ai/mcp"]
        }
      }
    }
    ```

    Or on **Pro/Max/Team/Enterprise** plans: **Settings → Connectors → Add** and enter `https://loadout.aident.ai/mcp`.
  </Accordion>

  <Accordion title="Cursor IDE">
    Add to `.cursor/mcp.json` in your project root:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="VS Code (Copilot)">
    Add to `.vscode/mcp.json` in your project root:

    ```json theme={null}
    {
      "servers": {
        "aident": {
          "type": "http",
          "url": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Windsurf">
    Add to `~/.codeium/windsurf/mcp_config.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "serverUrl": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Codex">
    Run in your terminal:

    ```bash theme={null}
    codex mcp add aident --url https://loadout.aident.ai/mcp
    ```

    Or add to `~/.codex/config.toml` (or `.codex/config.toml` for a trusted project):

    ```toml theme={null}
    [mcp_servers.aident]
    url = "https://loadout.aident.ai/mcp"
    ```

    Verify the server is configured:

    ```bash theme={null}
    codex mcp list
    ```
  </Accordion>

  <Accordion title="Gemini CLI">
    Add to `~/.gemini/settings.json`:

    ```json theme={null}
    {
      "mcpServers": {
        "aident": {
          "httpUrl": "https://loadout.aident.ai/mcp"
        }
      }
    }
    ```
  </Accordion>

  <Accordion title="Other Clients">
    Any MCP-compatible client (Goose, Kiro, OpenCode, Antigravity, Factory, etc.) can connect using the server URL `https://loadout.aident.ai/mcp`. Refer to your client's documentation for where to add MCP server configurations.
  </Accordion>
</AccordionGroup>

## Step 2: Authenticate

On first use, your MCP client will automatically:

1. Discover Aident's OAuth endpoints
2. Open your browser for login (Google OAuth or magic link)
3. Store your access token locally

No manual token management needed — the client handles everything.

## Step 3: Verify Connection

In your MCP client, try asking your AI assistant to use Aident tools. You should see Aident tools available across auth, skills, integrations, playbooks, templates, and dashboard.

## Optional: Migrate Local Integration Configs

After first-time setup, ask your agent to run `aident integrations migrate-local --json` if the host can run shell
commands. The command scans known local MCP/agent config files, redacts secret-like values, and shows which supported
integrations can be connected through Loadout. Connect all supported matches or a selected subset with
`aident integrations migrate-local --apply --integrationIds <ids> --json`, or use the MCP `vault` tool for the selected
Loadout integration IDs.

## Switching Accounts

To log out or switch to a different Aident account, use the `auth_logout` tool, then reconnect to trigger a new sign-in flow.

## Troubleshooting

### Authentication Issues

If the OAuth flow doesn't start automatically:

1. Restart your MCP client
2. Verify URL is exactly `https://loadout.aident.ai/mcp`
3. Check that your browser can reach loadout.aident.ai

#### "No OAuth flow is in progress" when running `complete_authentication`

Some MCP clients (notably Claude Code) expose two complementary tools — `authenticate` (start the flow) and `complete_authentication` (finish it by pasting the callback URL). The PKCE `code_verifier` produced by `authenticate` is held **in the client's process memory**. If that process restarts between the two tool calls (which can happen between message turns on long-lived agent sessions), the verifier is lost and `complete_authentication` returns `No OAuth flow is in progress`.

Reliable workarounds:

* **Open the authorization URL immediately** when prompted. The local loopback callback listener catches the redirect inside the same process, completing the flow before any state can be lost. This is the path that works without ever calling `complete_authentication`.
* If you must paste back the callback URL, run `authenticate` and `complete_authentication` back-to-back within the same agent turn.

Server-side, Aident emits the standard `WWW-Authenticate: Bearer resource_metadata=…` header on 401 responses, so MCP clients can rediscover the OAuth endpoints automatically.

#### Token TTLs

* Access token: 7 days
* Refresh token: 30 days (rotated on each refresh)
* Authorization code: 10 minutes

If your client supports automatic refresh, you should only see an interactive OAuth prompt every \~30 days. If you are re-authorizing more often, your client may not be persisting refresh tokens across sessions — check its token cache.

### "Skill requires unauthenticated integrations"

The skill needs integrations you haven't connected:

1. Use `vault` with `{ "action": "status" }` to see what's connected
2. Use `vault` with `{ "action": "connect", "integrationId": "<id>" }`
3. Authorize in browser
4. Retry the operation

### Connection Timeout

* Check firewall allows outbound HTTPS to loadout.aident.ai
* Configure proxy in MCP client if needed
* Try disconnecting VPN temporarily

### Tools Not Appearing

1. Restart MCP client
2. Verify URL in configuration file
3. Re-authenticate if token expired (client should handle automatically)

## Managing Your Connection

### Revoking Access

1. Open **Connections** from the left sidebar and go to **MCP**
2. Revoke the connection from your MCP client or Aident's MCP controls
3. All MCP connections will stop working immediately
4. Re-authenticate to reconnect

### Multiple Clients

You can use the same Aident account across multiple MCP clients. Each client maintains its own OAuth tokens.


# What is Aident AI?
Source: https://docs.aident.ai/documentation/overview

Build workflow automations by describing them in plain English

<Frame>
  <img alt="Your first automation, written and working." />
</Frame>

Aident AI lets you build workflow automations by describing them in plain English. Instead of wiring up flowcharts or writing code, you tell **Aiden** — the AI assistant in the editor — what you want done. Aiden turns it into a **Playbook** you can run on demand, on a schedule, or when something happens in another app.

## Who It's For

Aident is built for people who do the work, not the plumbing. If you can write a task out in a short paragraph, you can automate it.

<CardGroup>
  <Card title="Operators and ops leads" icon="user-gear">
    Recurring reports, handoffs, and check-ins that keep teams running.
  </Card>

  <Card title="Founders and small teams" icon="briefcase">
    Stitching together the handful of apps you already use, without hiring an engineer.
  </Card>

  <Card title="Sales, support, and marketing" icon="headset">
    Inbox triage, follow-ups, digests, and data entry that eat up your week.
  </Card>

  <Card title="Anyone tired of copy-paste" icon="clipboard">
    If the same steps show up every day, a playbook can take them over.
  </Card>
</CardGroup>

## What It Looks Like

When you open Aident, you'll see four main surfaces in the left sidebar.

<CardGroup>
  <Card title="Home" icon="house">
    The big input where you describe a new automation. This is how every playbook begins.
  </Card>

  <Card title="Dashboard" icon="grid-2" href="/documentation/triggers">
    All your playbooks, their current status, active trigger state, and last run.
  </Card>

  <Card title="Capabilities" icon="bolt">
    Browse and connect the integrations that power your agents and playbooks.
  </Card>

  <Card title="Connections" icon="plug">
    Manage MCP access, reusable skills, channels, and sandbox-related setup.
  </Card>
</CardGroup>

Your credit balance and the **Upgrade** button sit in the user center below the sidebar, so you always know where you stand.

## The Playbook

Every automation you build is a Playbook. Open one and you'll see three stacked sections on the canvas.

<CardGroup>
  <Card title="🎯 Goal" icon="bullseye">
    What you want done, and how you'll know it worked.
  </Card>

  <Card title="⚙️ Integrations" icon="plug">
    The apps your playbook needs — auto-discovered from your Steps.
  </Card>

  <Card title="📄 Plan" icon="list-check">
    An **Input** card, the ordered **Steps**, and an **Output** card.
  </Card>
</CardGroup>

Aiden drafts the first version from your description. You edit it directly, or ask Aiden in the side chat to rework anything. See [Playbooks](/documentation/playbooks) for the full tour.

## What You Can Automate

<AccordionGroup>
  <Accordion title="Email and communication">
    Summarize daily emails into a Slack digest, auto-reply to common questions, or route important messages to the right person.
  </Accordion>

  <Accordion title="Reporting and data">
    Pull numbers from a few tools into a weekly report, keep a spreadsheet updated, or watch a metric and alert when it crosses a line.
  </Accordion>

  <Accordion title="Tasks and projects">
    Turn Slack messages or emails into tasks, keep project status in sync across tools, and nudge follow-ups on a schedule.
  </Accordion>

  <Accordion title="Research and content">
    Research a topic and compile a summary, draft content from a template, or tag and file incoming information.
  </Accordion>
</AccordionGroup>

<Tip>
  A good starting description sounds like: "Every morning at 8am, check my Gmail for emails from clients, summarize them, and post the digest in #updates on Slack."
</Tip>

## Next Steps

<CardGroup>
  <Card title="Get Started" icon="rocket" href="/get-started">
    Create your first automation in minutes
  </Card>

  <Card title="How It Works" icon="diagram-project" href="/documentation/how-it-works">
    See what happens from Home to a running playbook
  </Card>
</CardGroup>


# Playbooks
Source: https://docs.aident.ai/documentation/playbooks

The canvas where you design every automation

A **Playbook** is the canvas Aident uses to describe an automation. It has three sections, and all of them appear on screen at once when you open the editor. The shape is the same for every playbook, so once you've seen one, you've seen them all.

<Frame>
  <img alt="The playbook canvas with Goal, Integrations, and Plan sections visible at once" />
</Frame>

You shape a playbook by talking to **Aiden** in the side chat — describe what you want, and Aiden updates the relevant section. Each section below explains what lives there and what it controls.

## The Three Sections

<CardGroup>
  <Card title="Goal" icon="bullseye">
    What you want done, and what success looks like.
  </Card>

  <Card title="Integrations" icon="plug">
    The apps your playbook needs — Gmail, Slack, and the rest.
  </Card>

  <Card title="Plan" icon="list-check">
    The input it takes in, the steps it follows, and the output it returns.
  </Card>
</CardGroup>

The headings on the canvas read **🎯 Goal**, **⚙️ Integrations**, **📄 Plan**, stacked top-to-bottom.

***

## 🎯 Goal

The **Goal** section is a short write-up of what the playbook is for — the problem it solves, the scope it covers, and how you'll know it worked. It's the source of truth Aiden uses when drafting and refining everything else.

A complete Goal usually expresses:

* **What should happen** — the outcome the playbook is responsible for.
* **When it's done** — what success looks like.
* **Where the result goes** — who sees it, and in which channel or format.

<Tip>
  When you ask Aiden to start a new playbook, the description you type becomes the first draft of the Goal. Sharper descriptions produce sharper goals.
</Tip>

***

## ⚙️ Integrations

The **Integrations** section lists every app the playbook depends on, shown as **integration pills**. Each pill carries a color that signals its readiness.

<AccordionGroup>
  <Accordion title="White pill — ready to use">
    Connected and authenticated. The playbook can use this app right away.
  </Accordion>

  <Accordion title="Orange pill — needs connecting">
    The app hasn't been signed in to yet. Until it's connected, any step that depends on it can't run.
  </Accordion>

  <Accordion title="Red pill — something's wrong">
    Aiden couldn't resolve this app for one of the steps. Asking Aiden to fix it from the side chat repairs the reference.
  </Accordion>
</AccordionGroup>

Integrations aren't picked manually. Aiden infers them from the Goal and the Steps in your Plan, and the pill list updates whenever the playbook changes.

***

## 📄 Plan

The **Plan** section is the executable part of the playbook. It's split into three pieces — Input, Steps, and Output — that together describe how a single run flows from beginning to end.

### Input

The **Input** card at the top of Plan describes what the playbook expects to receive each time it runs. Inputs come from a manual run, a trigger event, or a scheduled run.

There are two shapes:

* **Freeform** — a single text prompt.
* **Fields** — named pieces of data with types: **String**, **Number**, **Boolean**, **Array**, or **Object**.

The number on the Input card reflects how many fields are configured. Clicking the card opens the **Input** drawer, where you switch between Freeform and Fields and define each field's name, description, type, and whether it's required.

### Steps

The **Steps** are the ordered list of actions the playbook follows. Each step is a line of plain English that describes one thing the playbook does, in the order it does it.

Two things show up inside a step:

* **Skills** — the named actions Aident knows how to perform, shown as pills inline in the step.
* **Input fields** — if you've defined any, steps reference them by name so values flow through.

Aiden drafts the Steps from your Goal and refines them as the Goal evolves. Asking Aiden to add, remove, reorder, or reword a step from the side chat is the normal way to change them.

### Output

The **Output** card at the bottom of Plan describes the shape of what the playbook returns when it finishes. The shape determines how the result appears on the Dashboard and how downstream tools or playbooks consume it.

There are five shapes:

* **Freeform** — plain text. Best for summaries, drafts, and reports.
* **Markdown** — formatted text. Best for document-style results.
* **Web Page** — HTML output. Best when the result should render as a page.
* **JSON** — named fields with types. Best when another tool or playbook will read the result.
* **CSV** — rows and columns. Best for lists you'll open in a spreadsheet.

For JSON output, each field has a name and a type: **String**, **Number**, **Boolean**, **Array**, or **Object**. For CSV output, each column has a name and a type: **String**, **Number**, **Boolean**, or **Date**.

***

## Working with Aiden

The chat panel beside the canvas is how you change a playbook. Asking Aiden to "tighten the Goal", "add a step that posts to Slack", or "switch from Gmail to Outlook" updates the right section automatically. The canvas stays read-only while Aiden is editing — that's how you know the change is in flight.

***

## Next Steps

<CardGroup>
  <Card title="Triggers" icon="clock" href="/documentation/triggers">
    Configure when your playbook runs and what it takes in
  </Card>

  <Card title="Connect Your Tools" icon="plug" href="/documentation/integrations">
    Sign in to the apps your playbook uses
  </Card>
</CardGroup>


# Skills
Source: https://docs.aident.ai/documentation/skills

The named actions a playbook step can take

A **Skill** is one named thing Aident knows how to do — send an email, post a message in a channel, read a spreadsheet, search the web. Every step in a playbook either uses a skill or describes an action that maps to one.

***

## How a Skill Shows Up

Inside a step, a skill appears as a small pill on the line where it's used. The pill is a quick, at-a-glance reminder of what that step actually does.

<Frame>
  <img alt="A skill pill shown inline on a step in the playbook editor" />
</Frame>

* Hover over a skill pill to see what it does. Click it to open the related integration when there is one.
* If the skill requires an app, the matching pill in the editor's **⚙️ Integrations** section lights up.

You don't pick skills from a menu. Write your step in plain English and Aiden figures out which skill to reach for.

<Info>
  When Aiden drafts or edits your plan, it automatically finds skills that match what each step is trying to do. You can always rework a step and Aiden will re-resolve the skills.
</Info>

***

## Two Flavors

<CardGroup>
  <Card title="One-step actions" icon="bolt">
    Do a single thing, like sending an email or posting a message. Fast and focused.
  </Card>

  <Card title="Multi-step workflows" icon="layer-group">
    Bundle several actions into one reusable unit — for example, "research a topic and write a summary". One pill, several things happening under the hood.
  </Card>
</CardGroup>

Both flavors look the same when you use them in a step. The difference only matters when you're curious what's inside.

***

## Skills That Need an App

A skill that talks to the outside world needs an integration to work. "Send an email" needs Gmail. "Post a message" needs Slack, Discord, or Telegram. "Read a spreadsheet" needs Google Sheets.

When a step uses one of these skills:

<Steps>
  <Step title="Aiden adds the app">
    The needed integration shows up as a pill in the **⚙️ Integrations** section of your playbook.
  </Step>

  <Step title="You sign in once">
    If the pill is orange, click it and finish a standard sign-in. The pill turns white.
  </Step>

  <Step title="The skill just works">
    From there on, any step that uses this skill runs without asking you again.
  </Step>
</Steps>

See [Integrations](/documentation/integrations) for the full connection story.

***

## Browsing the Catalog

You don't have to browse skills to use them, but you can. Open **Connections** from the left sidebar and pick the **Skills** tab to see what Aident knows how to do across communication, productivity, research, developer tools, and more.

<Tip>
  Browsing the catalog is a nice way to get ideas for new playbooks. If you see something useful, write a Goal that describes it and let Aiden take it from there.
</Tip>

***

## Next Steps

<CardGroup>
  <Card title="Integrations" icon="plug" href="/documentation/integrations">
    Connect the apps your skills rely on
  </Card>

  <Card title="Playbooks" icon="book" href="/documentation/playbooks">
    See skill pills living inside real steps
  </Card>
</CardGroup>


# Triggers
Source: https://docs.aident.ai/documentation/triggers

Decide when your playbook runs

The **Triggers** tab is where you decide when a playbook runs without your hand on it. A playbook can have any number of triggers, and each one fires the playbook independently.

<Frame>
  <img alt="The Triggers tab on the playbook page" />
</Frame>

You don't need a trigger to run a playbook — you can always start a manual run from the button in the top right of the page. Triggers exist for the runs you don't want to launch yourself.

## Opening the Triggers Tab

From the [Playbook editor](/documentation/playbooks), click the **Triggers** tab at the top of the page. The header reads **Triggers**. Aiden stays available in the side chat — describe what you want and Aiden can configure triggers for you.

## Two Kinds of Triggers

<CardGroup>
  <Card title="Schedule" icon="clock">
    Run on a recurring time. Best for digests, reports, and anything tied to the clock.
  </Card>

  <Card title="Event Trigger" icon="bolt">
    Run when something happens in a connected app — a new Gmail message, a new Slack mention, a new row in a sheet.
  </Card>
</CardGroup>

When the tab is empty, the page shows a "No trigger configured" card with two shortcuts: **Schedule** and **Event Trigger**. Pick one to add your first trigger.

<Tip>
  Without any trigger, the playbook still works — it just only runs when you hit the manual run button.
</Tip>

***

## Schedule

A **Schedule** trigger fires on a recurring time. Use it for things that should happen on the clock — every morning, every Monday, the first of the month.

You'll set a cadence (how often it repeats) and the time it should fire. Schedules use your account's timezone.

Common patterns:

* Every weekday at 9am — for daily digests
* Every Monday at 8am — for weekly summaries
* Every hour — for active monitoring
* The first of the month — for monthly reports

***

## Event Trigger

An **Event Trigger** fires when something happens in one of your connected apps. Use it for anything reactive — answer a customer when they email, log a row when a deal closes, draft a reply when a new message comes in.

Setting one up takes three pieces:

* **The integration** — the app that produces the event (Gmail, Slack, Stripe, etc.).
* **The event** — the specific thing that should fire the trigger (for example, "new email received").
* **Optional filters** — narrow the event to a subset (only emails from VIP clients, only messages in #support).

Event Triggers can pass the event's data straight into the playbook as its input, so you don't have to set up a separate input shape for the run. You can also leave a short note for Aiden to use as guidance on each fire.

***

## Managing Triggers

Once a playbook has triggers, the Triggers tab shows a status bar at the top with one chip per trigger.

* A blue status bar means at least one trigger is active.
* A gray status bar means no triggers are active.

Click an individual trigger chip to enable or disable that trigger. The master **Enable all triggers** / **Disable all triggers** switch on the right side controls the whole list.

The badge on the Triggers tab itself shows how many triggers are currently active — quick at-a-glance confirmation that the playbook is set up to run on its own.

<Note>
  When no triggers are active, the Dashboard shows the playbook as **Not Ready**. Turn at least one trigger on for the status to flip to **Healthy**.
</Note>

***

## Next Steps

<CardGroup>
  <Card title="Back to the Playbook" icon="book" href="/documentation/playbooks">
    Edit the goal and steps these triggers run
  </Card>

  <Card title="Connect Your Tools" icon="plug" href="/documentation/integrations">
    Sign in to the apps your event triggers listen to
  </Card>
</CardGroup>


# Get Started
Source: https://docs.aident.ai/get-started

Your first automation, written and working.

It takes about five minutes to get you setup your own AI automation end-to-end without a single line of code just by plain language. Open up [app.aident.ai](https://app.aident.ai) in another tab and follow along.

## Create your first automation

Aident turns a plain-English description into a **Playbook** — a reusable automation you can run on demand, on a schedule, or when something happens in another app. Here's the path from idea to running playbook.

<Steps>
  <Step title="Describe what you want">
    On **Home** in the left sidebar, you'll see a big input with the placeholder *"Describe your job and automation here to get started..."*. Type a short paragraph describing the job — what should happen, which apps are involved, and how you'll know it worked. Press **Enter** to send. You can attach up to 10 images if a screenshot helps explain.

    <Tip>
      **Good starting descriptions:**

      * "Every morning at 8am, summarize my client emails from Gmail and post the digest in #updates on Slack."
      * "When a new row is added to my Google Sheet, create a task in Linear and notify the owner in Teams."
    </Tip>

    <Frame>
      <img alt="The Home input where you describe your automation" />

      <img alt="The Home input where you describe your automation" />
    </Frame>
  </Step>

  <Step title="Meet your Playbook">
    Aiden — the AI assistant — drafts a Playbook and opens it in the editor. You'll see three stacked sections on the canvas:

    * **🎯 Goal** — what the playbook should accomplish and what "done" looks like. Edit it like a rich-text field.
    * **⚙️ Integrations** — pills for the apps your playbook will use, inferred from your Goal and Plan.
    * **📄 Plan** — an **Input** card, the ordered **Steps**, and an **Output** card.

    Aiden sits in a side chat that's always available — ask it to reword your Goal, add a step, or rework any section. The initial draft can take a moment; hang tight while it finishes.
  </Step>

  <Step title="Connect the apps">
    Look at the integration pills in **⚙️ Integrations**:

    * **White pill** — connected and ready to use.
    * **Orange pill** — you haven't signed in yet. Click the pill to connect.
    * **Red pill** — something's unresolved. Click **Ask Agent to Fix** and Aiden will repair the reference.

    You can also manage app connections from **Capabilities** in the sidebar. **Connections** is for MCP access, reusable skills, channels, and sandbox-related setup.

    <Frame>
      <img alt="Connecting integrations from the playbook editor" />

      <img alt="Connecting integrations from the playbook editor" />
    </Frame>

    <Info>
      Don't see the app you need? [Request an integration](https://discord.gg/D9SuaZAQaJ) and we'll look into adding it.
    </Info>
  </Step>

  <Step title="Run it">
    Once all pills are white, it's time to try it out.

    * Click **Test Run** to trigger an on-demand run right now. Use this to confirm the Plan does what you expect.
    * Open the **Triggers** tab to configure automatic runs: add a **Time Trigger** for schedules or an **External Event** trigger for events in connected apps, such as a new email or a new row in a sheet. You can also set the **Input** it receives and the **Output** it returns.

    <Frame>
      <img alt="Test Run button in the playbook editor" />

      <img alt="Test Run button in the playbook editor" />
    </Frame>
  </Step>

  <Step title="Watch it on the Dashboard">
    Open **Dashboard** in the sidebar to see every playbook you own. A status label next to each one tells you what's going on:

    <AccordionGroup>
      <Accordion title="Running" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-running.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=f9cab42bce34182917c7388345d539db">
        A run is in progress right now.
      </Accordion>

      <Accordion title="Healthy" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-healthy.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=47446e10038380b1ca7b1c57aca439f5">
        The steady state — the playbook is set up and recent runs succeeded.
      </Accordion>

      <Accordion title="Awaiting" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-awaiting.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=ce9fa24a37e36089866891bb4c538b3e">
        The playbook is paused waiting for you — usually an approval or a piece of input. Click in to respond.
      </Accordion>

      <Accordion title="Failed" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-failed.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=3b71ab053eb3f71ec9d08c1981fa4f53">
        The last run didn't finish successfully. Click in to see what happened.
      </Accordion>

      <Accordion title="Not Ready" icon="https://mintcdn.com/aidentai/crni0s_EV3xqJFqx/images/icons/status-not-ready.svg?fit=max&auto=format&n=crni0s_EV3xqJFqx&q=85&s=f6f40050db0c9b6078f10724c82ba2f9">
        No triggers are active yet. Open the **Triggers** tab to turn one on.
      </Accordion>
    </AccordionGroup>

    Click any playbook to see its run history, open awaiting approvals, or jump back into the editor.

    <Frame>
      <img alt="Dashboard showing playbook statuses" />
    </Frame>
  </Step>
</Steps>

<Tip>
  Need another automation? Click **New Playbook** in the sidebar to start fresh, or head back to **Home** and describe something new.
</Tip>

## FAQ

<AccordionGroup>
  <Accordion title="A playbook shows Failed — what now?">
    Open the playbook to see which step failed and why. Common fixes: reconnect an orange integration pill, click **Ask Agent to Fix** on a red pill, or ask Aiden in the side chat to repair the step. Then try another **Test Run**.
  </Accordion>

  <Accordion title="My scheduled run didn't trigger">
    Open the playbook and go to the **Triggers** tab. Confirm a **Time Trigger** or **External Event** is configured and active. If Dashboard shows **Not Ready**, there are no active triggers yet. You can also ask Aiden to set triggers up for you from the side chat.
  </Accordion>

  <Accordion title="The playbook needs my approval to keep going">
    That's the **Awaiting** status on the Dashboard. Click into the playbook to review the run, approve or edit the pending item, and let it continue.
  </Accordion>

  <Accordion title="The results aren't quite what I wanted">
    Edit **🎯 Goal** to sharpen what "done" means, tweak the **Steps** in **📄 Plan**, or ask Aiden in the side chat to adjust a specific part. Run **Test Run** again to check. You're in control — the playbook is yours to shape.
  </Accordion>

  <Accordion title="I hit something that looks like a bug">
    We're in active beta and want to hear about it. [File a bug in Discord](https://discord.gg/kb6wpECN9W) and we'll take a look.
  </Accordion>
</AccordionGroup>


# Introduction
Source: https://docs.aident.ai/index

Your first automation, written and working.

<div>
  <div>
    <img alt="Decorative background image." />

    <img alt="Decorative background image." />
  </div>

  <div>
    <h1>
      Meet Aident AI
    </h1>

    <div>
      Describe a job in plain English and Aiden turns it into a Playbook you can run on demand, on a schedule, or when something happens in another app.
    </div>

    <div>
      <HeroCard title="Get Started" description="Describe and create your first workflow" href="/get-started" />

      <HeroCard title="Changelog" description="See the latest of what we just shipped" href="/changelog" />
    </div>
  </div>
</div>


# Loadout Integrations
Source: https://docs.aident.ai/loadout/integrations

Connect and manage the apps your agents can use

Open **Integrations** in Aident Loadout to browse the app catalog and connect the accounts your agents should use.

## Connect an Integration

<Steps>
  <Step title="Open Integrations">
    Go to [loadout.aident.ai/integrations](https://loadout.aident.ai/integrations).
  </Step>

  <Step title="Find the provider">
    Search by app name, capability, or category.
  </Step>

  <Step title="Open details">
    Select the integration to review the available actions and connection state.
  </Step>

  <Step title="Connect">
    Use **Connect** to start OAuth, or follow the Vault prompt for credential-based integrations.
  </Step>
</Steps>

## Connection Types

<AccordionGroup>
  <Accordion title="OAuth">
    Sign in with the provider in your browser. Aident stores the resulting token and agents call actions through
    Loadout.
  </Accordion>

  <Accordion title="Vault credentials">
    Admin users store API keys or delegated credentials in Vault when a provider does not support OAuth for the needed
    action.
  </Accordion>

  <Accordion title="Aident-managed integrations">
    Some built-in integrations are available without a user-managed provider credential.
  </Accordion>
</AccordionGroup>

## Disconnect an Integration

Use **Disconnect** from the integration drawer or card when a provider supports revocation from Loadout.

<Warning>
  Disconnecting an integration can break active agents, scheduled work, or MCP clients that depend on that provider.
</Warning>

## What Agents See

After an integration is connected, agents can discover and call supported capabilities through Loadout. They receive the
action result and audit receipt, not raw provider credentials.

## Next Steps

<CardGroup>
  <Card title="Vault and Audit" icon="key" href="/loadout/vault-and-audit">
    Manage credentials and review action-call history.
  </Card>

  <Card title="Troubleshooting" icon="life-ring" href="/loadout/troubleshooting">
    Fix missing tools, expired auth, and connection issues.
  </Card>
</CardGroup>


# Loadout Overview
Source: https://docs.aident.ai/loadout/overview

Use Aident Loadout to give AI agents governed access to your apps

Aident Loadout gives AI agents the integrations, credentials, policies, and audit trail they need to complete work
without handling raw provider credentials directly.

Open Loadout at [loadout.aident.ai/home](https://loadout.aident.ai/home).

## What You Can Do

<CardGroup>
  <Card title="Equip agents" icon="wand-magic-sparkles" href="/loadout/setup-agents">
    Copy the setup prompt or connect an MCP-compatible client.
  </Card>

  <Card title="Connect integrations" icon="plug" href="/loadout/integrations">
    Authorize apps such as Gmail, GitHub, Slack, Notion, and Google Workspace.
  </Card>

  <Card title="Manage Vault credentials" icon="key" href="/loadout/vault-and-audit">
    Store delegated credentials for integrations that do not use OAuth.
  </Card>

  <Card title="Review audit receipts" icon="clipboard-list" href="/loadout/vault-and-audit">
    See which action ran, where it came from, and whether it succeeded.
  </Card>
</CardGroup>

## Main Tabs

<Tabs>
  <Tab title="Home">
    Start setup, copy agent prompts, open one-click agent install links, and review recent Loadout usage.
  </Tab>

  <Tab title="Integrations">
    Browse the catalog, search by provider or capability, and connect the apps your agents should use.
  </Tab>

  <Tab title="Vault">
    Admin users manage stored credentials for API-key style integrations.
  </Tab>

  <Tab title="Audit">
    Review recent Loadout action calls, statuses, sources, and credit usage.
  </Tab>
</Tabs>

## How Loadout Protects Access

* Agents call Loadout tools instead of receiving provider secrets.
* OAuth connections are authorized in the browser and can be revoked.
* Vault credentials are stored centrally and managed by admins.
* Audit records show what happened after an agent calls an integration action.

## Next Steps

<CardGroup>
  <Card title="Set up agents" icon="terminal" href="/loadout/setup-agents">
    Connect Codex, Claude, Cursor, ChatGPT, or another MCP-compatible client.
  </Card>

  <Card title="Troubleshooting" icon="life-ring" href="/loadout/troubleshooting">
    Fix auth, missing tools, and integration connection issues.
  </Card>
</CardGroup>


# Set Up Agents
Source: https://docs.aident.ai/loadout/setup-agents

Connect coding agents and chat apps to Aident Loadout

Use Loadout from a coding agent, chat app, or any MCP-compatible client. The Loadout app shows the current setup prompt
and MCP URL on the Home tab.

## Coding Agents

For Codex, Claude Code, Cursor, OpenCode, OpenClaw, HermesAgent, Antigravity, and VS Code-style agent clients, paste this
setup prompt into the agent:

```text theme={null}
Follow https://aident.ai/SETUP.md
```

The setup flow installs or updates the Aident skill, then guides the agent through Loadout authentication and integration
access.

<Tip>
  Use the one-click install buttons on the Loadout Home tab when your client supports deep links.
</Tip>

## Chat Apps

Loadout exposes a hosted MCP server for chat apps and MCP clients:

```text theme={null}
https://loadout.aident.ai/mcp
```

<Steps>
  <Step title="Open your client settings">
    In ChatGPT, open Apps settings and enable Developer Mode. In Claude Desktop, open Connectors.
  </Step>

  <Step title="Add the Loadout MCP server">
    Create a custom app or connector and paste `https://loadout.aident.ai/mcp`.
  </Step>

  <Step title="Authorize in the browser">
    Sign in to Aident when the OAuth window opens.
  </Step>

  <Step title="Use Loadout tools">
    Start a new conversation and enable the Aident Loadout connector or app for that conversation.
  </Step>
</Steps>

## Environment URLs

Use the production URL unless your Aident team asks you to test another environment.

| Environment       | MCP URL                                 |
| ----------------- | --------------------------------------- |
| Production        | `https://loadout.aident.ai/mcp`         |
| Staging           | `https://staging.loadout.aident.ai/mcp` |
| Release candidate | `https://rc.loadout.aident.ai/mcp`      |

## Verify Setup

Ask your agent to list available Aident tools or to check Loadout integration status. A working setup should show tools
for authentication, capability search, capability execution, and Vault or integration management.

## Related Docs

<CardGroup>
  <Card title="MCP Server Setup" icon="terminal" href="/documentation/mcp-server-setup">
    Configure additional MCP-compatible clients.
  </Card>

  <Card title="Integrations" icon="plug" href="/loadout/integrations">
    Connect the apps your agents should use.
  </Card>
</CardGroup>


# Loadout Troubleshooting
Source: https://docs.aident.ai/loadout/troubleshooting

Fix common setup, authentication, and integration issues

Use this guide when an agent cannot see Loadout tools, authentication fails, or an integration action does not run.

## Tools Do Not Appear

1. Confirm the MCP URL is `https://loadout.aident.ai/mcp`.
2. Restart the MCP client or agent.
3. Re-authenticate if the client prompts for sign-in.
4. Ask the agent to list available Aident tools again.

## Authentication Does Not Start

* Check that your browser can reach [loadout.aident.ai](https://loadout.aident.ai).
* Confirm the MCP client supports remote HTTP MCP servers.
* Restart the client if it cached a stale unauthenticated state.

### "No OAuth flow is in progress"

Some clients keep the OAuth verifier only in process memory. If the client restarts between starting and completing auth,
the callback cannot be matched.

Reliable fixes:

* Open the authorization URL immediately when prompted.
* If pasting a callback URL manually, run the authenticate and complete steps back-to-back in the same agent turn.

## Integration Requires Authentication

Ask the agent to check Loadout or Vault status, then connect the missing integration from
[loadout.aident.ai/integrations](https://loadout.aident.ai/integrations). For credential-based integrations, ask an admin
to update Vault.

## Action Fails After It Starts

Open [Audit](https://loadout.aident.ai/audit) and inspect the failed entry. Common causes are expired provider auth,
missing provider permissions, deleted Vault credentials, provider rate limits, or invalid action input.

## Switching Accounts

Use the client's Aident logout tool if available, then reconnect. If the client stores tokens locally, you may also need
to clear that client's MCP token cache.

## Still Blocked

Contact support from the Loadout profile menu or join the Aident Discord from the same menu. Include the client name,
MCP URL, integration name, and the approximate time of the failed action.


# Vault and Audit
Source: https://docs.aident.ai/loadout/vault-and-audit

Manage delegated credentials and review Loadout action history

Vault and Audit are the operational controls for Aident Loadout.

## Vault

Vault stores delegated credentials for integrations that use API keys or shared service credentials instead of OAuth.
Only admin users can manage Vault credentials.

<Steps>
  <Step title="Open Vault">
    Go to [loadout.aident.ai/vault](https://loadout.aident.ai/vault).
  </Step>

  <Step title="Select an integration">
    Choose the provider that needs a stored credential.
  </Step>

  <Step title="Add or rotate credentials">
    Enter the credential values required by the provider and save.
  </Step>

  <Step title="Confirm readiness">
    Return to Integrations or ask your agent to check Loadout status.
  </Step>
</Steps>

<Warning>
  Rotating or deleting a Vault credential affects every agent and action that depends on that integration.
</Warning>

## Audit

Audit shows recent Loadout action calls. Use it to confirm what ran and diagnose failures.

Each audit entry can include:

* Action name
* Integration account
* Source client
* Status
* Duration
* Credits used
* Created time

## Common Review Flow

<Steps>
  <Step title="Open Audit">
    Go to [loadout.aident.ai/audit](https://loadout.aident.ai/audit).
  </Step>

  <Step title="Choose the time range">
    Use the date filters to focus on the period you need.
  </Step>

  <Step title="Inspect failures">
    Look for failed statuses and identify whether the issue is auth, provider response, missing input, or policy.
  </Step>

  <Step title="Retry after fixing access">
    Reconnect the integration or update Vault credentials, then retry from the agent.
  </Step>
</Steps>

## Next Steps

<CardGroup>
  <Card title="Integrations" icon="plug" href="/loadout/integrations">
    Connect or disconnect provider accounts.
  </Card>

  <Card title="Troubleshooting" icon="life-ring" href="/loadout/troubleshooting">
    Resolve common Loadout setup and runtime issues.
  </Card>
</CardGroup>