agentleFS
flopperjflopperj/dotfiles-core/.claude/skills/review-context/SKILL.md

review-context

Skill3 starsChanged 10 days agoMITView on GitHub

Generate or edit a per-project llms.txt that tells Scout/Ranger what historically breaks in this project. Stored at ~/.claude/review-context/<project>/llms.txt — outside any working repo. Use when the user types /review-context, asks to 'set up review context for this repo', 'tell reviewers what to watch for', or 'create a review-context file'.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After a post-mortem or incident — encode what broke so reviewers watch for it

    line 14
  2. Distinctive:

    Before requesting a review — priming Scout/Ranger with project-specific context

    line 15
  3. Distinctive:

    When reviewers keep missing the same class of issue in this repo

    line 16
  4. Distinctive:

    If absent: offer to generate a starter from the template below.

    line 25
  5. Distinctive:

    If present: display the current content and ask which sections to update.

    line 26
  6. Very specific:

    Never at repo root (./llms.txt)

    line 72
  7. Very specific:

    Never at .claude/review-context.md inside the repo

    line 73
  8. Distinctive:

    Never anywhere inside the working repo tree

    line 74

≈ “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

Review Context

Generate or edit a per-project llms.txt at ~/.claude/review-context/<project>/llms.txt (outside any working repo) that informs Scout and Ranger what historically matters in this project.

When to Use

  • Starting work on a project and wanting reviewers to know its sharp edges
  • After a post-mortem or incident — encode what broke so reviewers watch for it
  • Before requesting a review — priming Scout/Ranger with project-specific context
  • When reviewers keep missing the same class of issue in this repo
  • Updating stale context after significant architectural changes

Workflow

  1. Resolve the project slug: project=$(basename "$(git rev-parse --show-toplevel 2>/dev/null || pwd)")
  2. Determine the target path: ~/.claude/review-context/$project/llms.txt
  3. Assert storage safety: the target path MUST start with $HOME/.claude/review-context/. If it does not, fail immediately with: "Storage path assertion failed — will not write outside ~/.claude/review-context/". Never write inside the working repo.
  4. Check for existing file:
    • If absent: offer to generate a starter from the template below. Ask the user to fill in each section or provide bullet points to seed it.
    • If present: display the current content and ask which sections to update.
  5. Write the file: mkdir -p "$(dirname "$target_path")" then write or update the file.
  6. Confirm: report the path written and remind the user that Scout/Ranger will load this automatically on the next review of this project.

Checklist

  • Project slug resolved via git rev-parse --show-toplevel
  • Target path starts with $HOME/.claude/review-context/ (assertion checked)
  • ~/.claude/review-context/$project/ directory created
  • At least one section has content (not all blank)
  • No file created inside the working repo
  • User confirmed the written path

File Format

The skill writes a file with this shape (sections are headings; each section has 3–5 bullet points):

# Review Context — <project>

## Key Review Domains
- <what this codebase does and what makes it hard to review correctly>
- <the layer or service that receives the most bugs>

## Cross-Language Consistency
- <if the project spans multiple languages, what contracts must stay in sync>

## Data Integrity
- <schema invariants, idempotency requirements, or ordering constraints>

## Security
- <known sensitive paths, authn/authz patterns, anything that's been a vuln before>

## Performance
- <hot paths, known N+1 patterns, SLA-sensitive queries>

## Architecture
- <the key abstractions, which boundaries matter, what the "grain" of the system is>

Sections with nothing to say can be omitted. A file with 3 populated sections is better than 6 half-filled ones.

Storage Rule

The file MUST live at ~/.claude/review-context/<project>/llms.txt.

  • Never at repo root (./llms.txt)
  • Never at .claude/review-context.md inside the repo
  • Never anywhere inside the working repo tree

This is the user's explicit decision: personal context lives outside the repo to prevent accidental staging. Teammates cannot see it; that trade-off is accepted.

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.