Everything in one place.
How to write a spec in the web app, connect your coding agent over MCP or the intentdocs CLI, and check what it built.
01
Getting Started
- 1.1What is IntentDocs?IntentDocs turns your idea into a spec your coding agent builds from, then checks the agent built what you meant.
- 1.2Creating a projectStart a project in the web app, answer the five vision questions, and publish the story map IntentDocs drafts from them.
- 1.3Connecting your agentPoint Claude Code, Cursor, Claude.ai or any MCP client at the hosted IntentDocs server so your agent reads the spec and reports back.
- 1.4Installing the CLIRun the intentdocs CLI with npx or install it globally, then sign in with intentdocs login.
- 1.5Connecting a repoConnect a repository to a project with npx intentdocs init, set up your coding agent, and link GitHub so pull requests and CI results reach each story.
- 1.6Brownfield agent promptThe prompt that has your coding agent read an existing codebase and write a report you import into IntentDocs, and the report format it follows.
02
Using the Web App
- 2.1A project in the web appHow a project is laid out in the web app, what the Overview shows, and how the phone and desktop layouts relate.
- 2.2The Spec tabWrite and shape the spec on the Spec tab, from the vision and story map to releases, filters, story cards, the data model and design.
- 2.3The story pageEverything about one story on one page, from the status card at the top to its Spec, Plan, Tests and Review tabs.
- 2.4The Plan tabPut stories in build order, see the phases agents can build in parallel, and start build runs for your agents to work through.
- 2.5The Validate tabReview what agents built, approve stories one at a time or in bulk, read how far to trust each one, and close releases up to Go live.
- 2.6Working with agentsHand work to agents from the web app, from the in-app assistant and copyable prompts to one-tap handoffs, claims, blockers and the rules for what agents may build.
- 2.7Workspaces and teamsPersonal and team workspaces, what each role can do, guests on a single project, who approves plans and stories, and how agents are bound to a workspace.
03
MCP Integration
- 3.1MCP overviewWhat the IntentDocs MCP server gives a coding agent, and the rules it builds by.
- 3.2Claude Code setupConnect Claude Code to the hosted IntentDocs MCP server, or run the local server through the CLI.
- 3.3Cursor & remote setupConnect Cursor, Claude, Grok Bot, and other MCP clients to IntentDocs.
- 3.4Tools & resourcesEvery tool, resource and prompt on the hosted and local MCP servers, with parameters, results and refusals.
- 3.5Example workflowsEnd-to-end tool sequences for building a story, building a run, importing an existing codebase and fixing drift.
- 3.6Spec my ideaThe steps a chat assistant follows from an app idea to a story map the founder reviews, and where it stops for the person.
04
CLI Reference
- 4.1CLI overviewEvery intentdocs command in one table, plus how the CLI signs in, names stories and writes files.
- 4.2Auth and keysSign in with intentdocs login, see which workspace your token reads, and save your own Anthropic key for AI features.
- 4.3Project setupCreate a project from the terminal, connect a repository to its story map, keep INTENT.md current and configure your coding agent.
- 4.4Story workflowThe daily loop from the terminal — see progress, pick up a story, run its criteria, mark it done, stop it or add a new one.
- 4.5Decisions and memoryLog, list, amend and retract decisions, add notes to a story, and add code examples your agent should follow.
- 4.6AI featuresScore story complexity, research a question with your project context, and turn a PRD into stories.
- 4.7Story mapsPlan a change as stories with propose, and move a whole story map in from an existing codebase with map extract and map import.
- 4.8IntegrationsRun the local MCP server, register it with OpenClaw, and report test results from CI to each story.
05
Core Concepts
- 5.1VisionThe five questions that ground a project, how they gate the story map, and how agents read and write them.
- 5.2Story maps & INTENT.mdHow a project's spec is structured, from activities and stories to releases and statuses, and how INTENT.md carries it into your repository.
- 5.3Acceptance criteriaWhat a criterion holds, how its priority and verification method work, how results are reported, and the gate that keeps a story from being marked dev complete too early.
- 5.4Build orderHow the build order, dependencies, phases and build runs decide which story an agent builds next, and what a person must do before it starts.
- 5.5VerificationWhere the evidence that a story works comes from, how CI results reach each story, and how IntentDocs tells you how far to trust a "done".