agentleFS
IgorGanapolskyIgorGanapolsky/mac-yolo-safeguards/.agents/skills/egm-fresh-review/SKILL.md

egm-fresh-review

Skill3 starsChanged 4 days agoMITView on GitHub

Refuse a Goldfish review that the same context graded, a child prompt that already contains the verdict, host instruction files in the child file list, a plugin install, or a vendor timing recorded as a local result. Require a different reviewer, ranked findings, and at most one re-audit before a fix or ship claim counts.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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:

    Use this before treating a fix, a ship claim, or a design as reviewed.

    line 8
  2. Very specific:

    Write the receipt only after the fresh reviewer returns.

    line 36

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

Fresh-context review

Use this before treating a fix, a ship claim, or a design as reviewed.

Source: https://github.com/google/elephant-goldfish-model

A review in the same context is not a Goldfish review. If this harness cannot spawn a fresh subagent, stop. Do not call the result a Goldfish review.

What this gate does

  • isolation must be fresh.
  • reviewerAgentId and parentSessionId must both be set, and they must differ.
  • mode is full (stages 5–9), feature_spec (5–8), implementation (9), or claim (9).
  • reauditCount is 0 or 1.
  • An open blocker or an open major fails the receipt. An open nit does not.
  • A childPrompt that already states the verdict fails as parent_verdict. The matchers live in tools/egm-fresh-review-gate.js.
  • AGENTS.md, CLAUDE.md, GEMINI.md, or .cursorrules in childContextFiles fails as host_instructions.
  • kind or action of install, or an install command for this plugin, fails as plugin_install. The gate does not install the plugin.
  • A note with measuredHere: true that repeats a vendor timing fails as vendor_figure. Those figures are not local results.
  • The gate does not call a model.

Commands

node tools/egm-fresh-review-gate.js
node tools/egm-fresh-review-gate.js --review receipt.json
node tools/egm-fresh-review-gate.js --install
node tests/test-egm-fresh-review-gate.js

Write the receipt only after the fresh reviewer returns. Put the reviewer's evidence in findings, not a paraphrase from the parent.

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.