Skip to content

Control Adoption

ContributorsMyles Henaghan

When to use: Deliberately invoke in Agent mode against an in-scope consumer repository to see which library deterministic controls it already has, and what it does better or newer that should be pulled back into the library. Re-run periodically (e.g. quarterly) to keep provenance honest.

Purpose

A dual-direction diagnose skill:

Direction Question
Adoption (library → repo) Which library controls does this repo already have?
Backport (repo → library) What does the repo do better or newer that should land in the library?

Instructions

  1. Read references/control-adoption-prompt.md and references/control-catalogue.md.
  2. Run the scope gate (target repo + archetype, control set, results path) before scanning. Read-only on the target repo; write only into the portfolio path that owns the matrix.
  3. On first use in a portfolio, seed control-matrix.yml and reviews/ from assets/ if those files do not exist yet.
  4. For each control: evidence-led status (present / partial / absent / N/A), direction (adoption / backport / parity), and concrete paths/flags.
  5. Update state (control-matrix.yml) and run log (reviews/YYYY-MM.yml) every run. Write the optional narrative only when asked.
  6. Report the net diff (or baseline), updated counts, and new backport candidates. Flag effort/provenance implications — do not silently rewrite effort plans.

What this scan covers

Artefact Job
Control matrix Current per-repo control status, since dates, provenance
Run log When each repo was scanned + per-run changes net diff
Narrative (optional) Human-readable evidence write-up with At-a-glance

References


Troubleshooting

Symptom Likely cause Fix
Distorted present/absent mix Control marked ⬜ when N/A for archetype Re-scope by archetype; use — for N/A
Vague "present" claims No file/flag cited Require path + tool/flag before ✅
Zero adoption but strong controls Independent implementation Treat as backport source; list extraction targets
Silent effort drift Matrix updated without reporting implications Report provenance/effort deltas; wait for user to apply
Matrix missing on first run Portfolio never seeded Copy assets/control-matrix.example.yml into the confirmed results path