rfc — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited rfc (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
<objective> Produce a complete RFC document via 10 disciplined micro-steps. Each step focuses on one phase, persists state in frontmatter, and never converges to a plan before alternatives + tradeoffs are explicit. </objective>
<when_to_use>
Don't use for: trivial refactors, single-file fixes, doc-only changes. Use ship or direct edit instead. </when_to_use>
<parameters>
<short title>: RFC title (required). Used to generate slug.--auto: skip AskUserQuestion confirmations between steps--scope <path>: limit codebase context gathering to a path (default: repo root)--no-review: skip step-08 adversarial review--out <dir>: output dir (default: docs/rfcs/ if git repo, else ~/.claude/rfcs/)</parameters>
<state_variables>
rfc_id: 4-digit zero-padded (0001, 0002, ...)rfc_slug: kebab-case from titlerfc_dir: absolute path <out>/NNNN-slug/rfc_path: <rfc_dir>/RFC.mdauto_mode: booleanscope_path: path for codebase explorationskip_review: booleanstatus: Draft | Review | Accepted | Rejectedcontext_collected: object (symbols, files, prior art from GitNexus/grep)problem, motivation: stringsalternatives: array of {name, summary, pros, cons}design: object (architecture, data model, API, modules touched)drawbacks, risks, open_questions: arraysrecommendation: string + rationaleimpl_plan: ordered task listreview_findings: array (from step-08)stepsCompleted: array (tracks progress in frontmatter)</state_variables>
<entry_point> Load steps/step-00-init.md </entry_point>
<step_files>
| Step | File | Purpose |
|---|---|---|
| 00 | steps/step-00-init.md | Parse args, create RFC dir, init frontmatter |
| 01 | steps/step-01-context.md | Gather codebase context (GitNexus + grep) |
| 02 | steps/step-02-problem.md | Define problem statement + motivation (interview) |
| 03 | steps/step-03-alternatives.md | Surface alternatives + tradeoffs |
| 04 | steps/step-04-design.md | Proposed design: modules, data, API |
| 05 | steps/step-05-risks.md | Drawbacks, risks, unknowns |
| 06 | steps/step-06-recommendation.md | Recommendation + rationale |
| 07 | steps/step-07-impl-plan.md | Implementation plan (ordered tasks) |
| 08 | steps/step-08-review.md | Adversarial review via subagent (optional) |
| 09 | steps/step-09-finalize.md | Finalize, set status, summarize |
</step_files>
<references>
| File | Content |
|---|---|
| references/rfc-template.md | Markdown template: 11 sections |
| references/interview-questions.md | Question bank per step (problem, alternatives, design) |
</references>
<integrations>
mcp__gitnexus__query, mcp__gitnexus__context, mcp__gitnexus__impact for codebase explorationmcp__exa__web_search_exa, mcp__exa__get_code_context_exa for prior art / industry patternsAgent(subagent_type=general-purpose) for adversarial review (step-08)</integrations>
<critical> RFC ≠ Plan. NEVER converge to implementation before alternatives + tradeoffs are written. Plan mode = biased toward action. RFC slows agent down in the right place. If user says "just plan it", redirect to ship skill: RFC is for boundary-crossing decisions only. </critical>
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.