agentleFS
mgiovanimgiovani/cc-arsenal/skills/docs-rfc/SKILL.md

docs-rfc

Skill8 starsChanged 48 days agoMITView on GitHub

Create a numbered RFC (Request For Comments) document proposing a

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Create a new RFC document proposing and discussing a change.

    line 19
  2. Distinctive:

    Verify current state: explore the codebase to understand what exists today.

    line 25
  3. Distinctive:

    Check dependencies: confirm libraries/tools mentioned actually exist in the project.

    line 27
  4. Very specific:

    No Task tool available: explore directly with grep/glob/read before writing, covering the same four questions (current patterns, related …

    line 43
  5. Distinctive:

    Every hit is an unfilled placeholder or an unfinished draft: draft the content and re-run.

    line 78
  6. Distinctive:

    Write the RFC before starting implementation, not after

    line 103
  7. Distinctive:

    Document alternatives considered and why they were rejected

    line 105
  8. Distinctive:

    Keep it updated as a living document during review

    line 107

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Create request for comments

Create a new RFC document proposing and discussing a change.

Anti-hallucination guidelines

RFCs propose changes to real systems, so ground every claim before writing:

  1. Verify current state: explore the codebase to understand what exists today.
  2. Reference actual code: don't invent APIs or patterns, find real examples.
  3. Check dependencies: confirm libraries/tools mentioned actually exist in the project.
  4. Validate assumptions: each claim about current state must be verified.

Workflow

Phase 1: explore and gather context

Understand the codebase before writing anything. If the Task tool is available, use the Explore agent:

Use Task tool with Explore agent:
- prompt: "Analyze the codebase to understand [RFC_TOPIC]. Find: 1) Current implementation patterns, 2) Related components and their interactions, 3) Existing similar features, 4) Technical constraints. Return verified findings with file paths."
- subagent_type: "Explore"

No Task tool available: explore directly with grep/glob/read before writing, covering the same four questions (current patterns, related components, existing similar features, technical constraints). Either way, keep what you find: it feeds the Background and Detailed Design sections in Phase 2.

Phase 2: fill the template

Copy this checklist and tick each step:

- [ ] Parse `$ARGUMENTS`: title, plus an optional variant keyword (`minimal`, `standard`, `detailed`); remove the keyword from the title; default `standard`
- [ ] Number: scan `docs/rfc/` for `RFC-XXXX-*`, add 1 to the highest (start at `0001`), 4-digit padded
- [ ] Slug: kebab-case the title, lowercase, special characters stripped ("Add GraphQL API Support" -> `add-graphql-api-support`)
- [ ] Author: `git config user.name`, falling back to `"Development Team"` if empty
- [ ] Load the variant template from `<skill-dir>/assets/templates/` and draft real content for every `{{PLACEHOLDER}}` from the Phase 1 findings or explicit reasoning
- [ ] Write `docs/rfc/RFC-XXXX-kebab-case-title.md` (create `docs/rfc/` if missing) with status "Draft"
- [ ] Run the leftover-token check below; fix and re-run until it prints nothing

Variants:

  • minimal -> minimal.md: Summary, Motivation, Proposal, Open Questions. Use for small changes.
  • standard -> standard.md (default): adds Rationale and Alternatives, Implementation Plan, Testing Plan, Migration Strategy, Timeline. Use for most feature proposals.
  • detailed -> detailed.md: full set including Goals/Non-Goals, Security Considerations, Performance Implications, Monitoring and Metrics. Use for major/architectural changes.

Each variant has its own placeholder set (metadata fields, body sections, risk tables, alternatives, review history, and so on). Never leave a placeholder token literally in the output.

Leftover-token check, run on the file just written:

grep -nE '\{\{|\}\}|TODO|TBD' docs/rfc/RFC-XXXX-kebab-case-title.md

Every hit is an unfilled placeholder or an unfinished draft: draft the content and re-run. After 3 rounds, stop and report what still matches.

Phase 3: report creation

Show the RFC number, title, file path, and next-step guidance (share for feedback, update status as it progresses). This is the last step; nothing downstream consumes this report beyond the person reading it.

Usage examples

docs-rfc "Add GraphQL API Support"
docs-rfc minimal "Update Logging Format"
docs-rfc detailed "Migration to Microservices Architecture"

RFC status lifecycle

Draft -> In Review -> Accepted / Rejected (or Withdrawn at any point, Implemented after accepted work ships). Update the status field as the RFC moves through review.

Good practices

  • Write the RFC before starting implementation, not after
  • Include concrete examples and real code references, not hypotheticals
  • Document alternatives considered and why they were rejected
  • Link related ADRs, issues, or other RFCs
  • Keep it updated as a living document during review

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.