.agents/skills/rfc-review/SKILL.md
Review a draft RFC for scope, clarity, and contributor-readiness. Use before opening an RFC for comment or merging a proposal branch. Produces concrete suggested revisions, not just a report.
npx skillsauth add akollegger/aie-matrix rfc-reviewInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Use this skill to evaluate a draft RFC against a consistent set of quality criteria before it is opened for comment. Works for self-review by the author, maintainer review during a PR, or peer feedback from a collaborator. Produces specific, actionable suggested revisions — not a checklist of observations.
proposals/rfc/.proposals/rfc/README.md to confirm structural compliance.Is the RFC the right size for its stated purpose?
Flag: RFC specifies code, method signatures, or implementation detail that belongs in the implementation, not the proposal.
Flag: RFC is so vague that an implementer would face constant ambiguity about what was agreed.
Can a contributor tell when the work is done by observing something, not by checking off packages created?
Flag: No user stories, demo scenario, or observable acceptance criteria present.
Flag: Acceptance criteria describe structure ("the registry package exists") rather than behavior ("a ghost can register and receive a token").
Is there a concrete path a contributor follows to verify the work?
Flag: No demo scenario or "getting started" path described.
Flag: Demo scenario assumes knowledge not present in the RFC or repo README.
Is there enough information for implementation to begin without another round of design?
Flag: A contributor would need to make significant design decisions not acknowledged as open questions.
Flag: Open questions are present but not listed — they are buried in prose or implicit.
Is the RFC constraining implementation detail that should be left to the implementer?
Flag: Code blocks present that specify implementation rather than illustrate intent.
Flag: RFC would be wrong if an implementer chose a different variable name or data structure.
Are the alternatives realistic comparisons that help a reader understand why this approach was chosen?
Flag: Alternatives are strawmen that no contributor would actually propose.
Flag: A significant alternative is missing — one a reviewer would ask "why not X?" about.
Does the RFC align with decided components and open questions in docs/architecture.md?
Flag: RFC introduces a technology or pattern inconsistent with docs/architecture.md without explanation.
## RFC Review: [RFC title]
### Summary
[2-3 sentences: overall assessment and whether it is ready to open for comment]
### Blocking Issues
[Issues that must be resolved before the RFC is ready. For each: criterion, problem, suggested revision.]
### Advisory Issues
[Issues worth addressing but not blocking. Same format.]
### Strengths
[What is working well and should be preserved.]
For each blocking or advisory issue, include:
data-ai
Convert existing tasks into actionable, dependency-ordered GitHub issues for the feature based on available design artifacts.
data-ai
Generate an actionable, dependency-ordered tasks.md for the feature based on available design artifacts.
testing
Create or update the feature specification from a natural language feature description.
content-media
Execute the implementation planning workflow using the plan template to generate design artifacts.