OpenStory

Use when asked to give a narrated/driven tour of the OpenStory dashboard, tell a "user story" from session data, demo features by driving the UI, tune the design live with voice narration, or record a tour video. Drives the UI through the control seam while narrating aloud.

Query your own coding sessions via the OpenStory MCP tools — find past work, understand what happened, check token usage, search across history, and watch live sessions stream. All read-only observability, never interferes with your execution.

Run OpenStory's automated red-team — dependency scanners, aggressive in-process tests, and testcontainer-based exploit probes — then report which attack vectors were fended off and which broke through. Use when the user asks "audit security", "run a red team", "are we secure", "what attacks would work", or before merging anything that touches auth, the HTTP surface, or dependencies. Dogfoods the project's own tooling — never invent ad-hoc checks when the runner exists.

Tell the story of a team's work on a given day by running the deterministic team_day pipeline, reading the resulting facts sheet, and composing a short narrative on top. Use when the user asks "what did our team do today / yesterday / on date X", "summarize the team's day", "what did Max and Katie work on", or wants a per-author day report grounded in OpenStory data. Dogfoods OpenStory — never guess from the conversation, always run the pipeline first.

Query OpenStory for a structural report on an agent session — at-a-glance metrics, turn arc, file impact, errors, and optional event detail. Use when you have a session ID and need to know what happened in it without re-reading raw transcripts.

Validate the project's docs against the live codebase. Catches drift between claims (in markdown prose) and reality (in Cargo.toml workspace members, source files, the filesystem). Use before committing docs changes, or whenever you're about to write something architectural and want to be sure you're not perpetuating a stale claim. The validator surfaces things internal consistency cannot catch — when every doc agrees with every other doc but none agree with the build.

Tell the story of an OpenStory session by collecting deterministic facts (records, tools, patterns, prompts) via the REST API, then narrating them. Use when the user asks "what happened in session X", "summarize this session", "what was that session about", or wants to understand a past or current session through OpenStory's own data. Dogfoods OpenStory — never grep transcript files when you can ask the API.