CLI Reference

AI features

intent complexity, research, and import — AI-powered commands gated to the Builder plan or unlockable with your own Anthropic key.

These three commands call Claude to do real analysis work, using your project's context. They're gated to the Builder plan; on a lower plan the API returns a 402 and the CLI prints an upgrade message pointing at intentdocs.com/pricing. You can unlock all of them on any plan by setting your own Anthropic API key — see intent keys set.

intent complexity [storyId]

intent complexity                # analyze every non-done story
intent complexity a1b2c3         # analyze one story

Scores complexity on a 1–10 scale with reasoning, risks, and a suggested effort estimate for each story, sorted highest-complexity first with a bar chart. Prints a summary: average complexity and how many stories scored as high-complexity. Useful before starting a release, to catch a story whose effort: M estimate is hiding real complexity.

intent research [query]

intent research
intent research "What are the tradeoffs between JWT and session-based auth for this app?"

Sends your query to Claude along with your project's context and streams back a markdown research write-up. Prompts for the query if you don't pass one.

intent import [file]

intent import prd.md
intent import              # reads from stdin

Parses a PRD or spec document into structured stories. Reads from a file if given, otherwise from stdin. Sends the text (format: PRD, Markdown, or plain text) for parsing, which returns a hierarchy of activities → steps → stories with acceptance criteria already extracted. Shows you the parsed hierarchy and asks for confirmation before creating anything — nothing is written to your story map until you approve.