Skip to content

Goldfish Review Resolver

Triages reviewer findings on a document and proposes or applies fixes to it

Stars
9,965
Licence
Apache-2.0
Files
2 files

What it does

This skill lets an agent take a list of problems flagged by reviewers of a document (the Goldfish review format used in the Elephant Goldfish model project) and sort each one into a category: contest it with evidence, escalate it to the user, propose a text fix, or decline it as a minor nit. It writes a markdown report of these decisions, then pauses and waits for you to choose: auto-accept the proposed fixes, skip them, or walk through each item one by one. Only when you accept does it edit the target document.

Great for

  • Authors who received structured review findings on a design or technical document and want a first-pass triage
  • Teams running the Elephant/Goldfish multi-reviewer workflow who want consistent resolution of flagged items

Skip it if

  • General proofreading or editing of a document with no reviewer findings; the skill expects a specific findings list format
  • Auto-accepting fixes on important documents without reading the report, since the agent may misjudge a finding and edit the file directly

Try asking

  • “Resolve the findings from my Goldfish review of design/spec.md lines 40-120 and write the report”
  • “Walk me through each proposed fix interactively before changing anything”
  • “Continue to the next phase without applying any fixes, just give me the report”
Step-by-step guide: how to use Goldfish Review ResolverInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 2 original files, the Apache-2.0 licence and an attribution note.

    egm-resolve.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip egm-resolve.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip egm-resolve.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run it in the same conversation session that produced the document so the agent can recall why choices were made, as the skill recommends.
  • Read goldfish_review_report.md before replying 'auto accept', especially for ESCALATE and CONTEST items, which are excluded from auto-accept but need your judgement.
  • Use 'interactive breakdown' for high-stakes documents so each proposed edit is explained before it is applied.

What's inside

  • agents/openai.yaml
  • SKILL.md

Redistributed unmodified from google/elephant-goldfish-model under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-11-02. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.