Run an assessment¶
Prerequisites¶
- The plugin installed in Copilot CLI or Claude Code.
- Your terminal / tool open with the target repository as the working directory, so the scan can see its files.
Run it¶
Use the slash command:
Or trigger it by natural language — the skill responds to phrases like:
assess our AI readiness · where are we on the framework? · check our habitat maturity · score our AI maturity
What happens¶
- Discovery — a report of the habitat documents and signals found (with paths) and the ones absent.
- Questions — 3–5 clarifying questions, asked one at a time, to place the behavioural dimensions the filesystem can't show. Answer each before the next is asked.
- Assessment — all fourteen model dimensions placed L1–L5, the cognitive level, and the Habitat Build Gap.
- Report — written to
assessments/YYYY-MM-DD-assessment.md, with a short summary in the chat.
See the full assessment output structure for a section-by-section breakdown.
Tips¶
- Run it from the repo root. Running from a subdirectory hides habitat documents and skews the result.
- If two files seem to fill the same role (say two candidate instruction files), the assessment will stop and ask which is canonical rather than guess. Answer it — silent picks produce confidently-wrong assessments.
- Commit the report.
assessments/YYYY-MM-DD-assessment.mdis a durable record; committing it lets you track movement over time.
Next¶
- Read the Habitat Build Gap to act on the result.
- Render the HTML report to share it.
- Want rigorous per-dimension scoring? Run the precise survey.