AgentIndex icon
AgentIndex
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Submit Tool
ToolsMCP ServersHOTCategoriesTrendingNewBest ListsCompareGuides
Home/
Dev Tooling/
elephant-goldfish-model
elephant-goldfish-model logo

elephant-goldfish-model

Active·★ 3.1k·Apache-2.0·Updated 2026-10-27

Elephant-Goldfish Model (EGM) is an Apache-2.0 plugin from Google for designing features with coding agents before any code is written. The README says thousands of Google engineers use it. The Elephant agent holds your project context and drafts the specification; Goldfish reviewers read only the draft and the source it references, without the earlier conversation, so unstated assumptions and blind spots surface. You resolve the trade-offs and approve the design.

It installs as a plugin in Antigravity, Claude Code and Codex, with setup notes for Claude Cowork and ChatGPT Work. A typical run is egm-brainstorm to agree a direction in concepts_brief.md, egm-draft to write design_draft.md section by section, and egm-review to have fresh reviewers check clarity, fit with the code and the implementation plan; each skill can also be used alone. Reviews need a harness that can spawn fresh subagents, and EGM stops rather than fake one. The README notes it is not an officially supported Google product and does not replace tests.

This write-up is based on the project's README and public repository data. We have not hand-tested the tool.

#design-docs#spec-review#claude-code#codex#antigravity#subagents#agent-skills
↗ Visit site★ GitHub
01

Features

01Elephant agent drafts the spec with full project context
02Goldfish reviewers check the draft without the earlier conversation
03Brainstorm, draft and review skills usable on their own
04Plugin installs for Antigravity, Claude Code and Codex
05Stops the review if fresh subagents are unavailable
02

Why choose it

+Reviewers see only the draft and its sources, so they catch assumptions the drafting agent took for granted
+Installs as a plugin in Claude Code, Codex and Antigravity
+Brainstorm, draft and review skills can be used separately
03

Trade-offs

!Not an officially supported Google product.
!Needs a harness that can create fresh subagents; otherwise reviews stop.
!Overkill for small, clear bug fixes, as the README itself says.
04

Maintenance check

ActiveChecked: 2026-10-27
Last commit
2026-10-27
Commits, last 90 days
98
Releases, last 90 days
1
Latest release
v0.1.0 (2026-10-27)
Open issues
45
Recent issues with at least one comment
3/13

Status rule: archived on GitHub = Archived; last commit within 90 days = Active; 91 to 365 days = Slowing; older = Stale. All figures come from the GitHub API on the date shown.

Most discussed open issues

  • #110 Store EGM outputs in a dedicated folder3 comments · 2026-10-22
  • #146 Add verified OpenCode support2 comments · 2026-10-24
  • #186 [Feature]: Add support for OhMyPi (omp)1 comments · 2026-10-27
  • #175 [Feature]: Add /ask-egm command for contextual help and guidance1 comments · 2026-10-26
  • #76 Add an optional whiteboard check before spec verification1 comments · 2026-10-08

Issue titles are quoted from GitHub. Open the link to check whether a problem still applies to the version you use.

05

Use cases

↳Writing a design doc before building a feature with an agent
↳Finding hidden assumptions in a technical spec
↳Standardizing design reviews across a team using coding agents
06

Alternatives

⚡Top VS:
VSoh-my-hermes★4.4k→VSawesome-gamedev-agent-skills★5.1k→VStotally-integrated-claude★75→VSkungfu★4.5k→
oh-my-hermes logo
oh-my-hermes★ 4.4k
The packages for Hermes Agent ⚚ skills, coding harness, long-term memory, model optimized subagent & bot team workflows
VSCompare→
awesome-gamedev-agent-skills logo
awesome-gamedev-agent-skills★ 5.1k
74 game-dev skills for AI coding agents — Godot, Unity, Unreal, Phaser, PixiJS, three.js, Bevy, pygame, LÖVE, Roblox. Portable SKILL.md Agent Skills (the format Anthropic launched as Claude Skills), with a router that loads the right skill for your engine and task. Runs in Claude Code, Cursor, Kiro, Codex, Copilot, Gemini CLI and more.
VSCompare→
totally-integrated-claude logo
totally-integrated-claude★ 75
A Claude Code plugin for Siemens TIA Portal engineering automation.
VSCompare→
kungfu logo
kungfu★ 4.5k
Your agents don’t hand off the work. Kungfu keeps the same Work moving across Codex, Claude, OpenCode, and your own execution surface.
VSCompare→
zeroshot logo
zeroshot★ 2.0k
Open-source meta-harness for coding agents. Define your own workflows, reviewers and repair loops. zeroshot enforces the process you configure.
VSCompare→
ai-devkit logo
ai-devkit★ 1.7k
The control plane for AI coding agents.
VSCompare→
Lody logo
Lody★ 1.6k
Share coding agents with your team on phone and desktop
VSCompare→
reticle logo
reticle★ 1.3k
AI agents can generate code, but still struggle to understand what they build. Reticle brings Jev-style machine-native runtime perception to web & desktop applications.
VSCompare→
See all alternatives →

Comments

Log in to leave a comment

No comments yet. Be the first!

On this page
01Features02Why choose it03Trade-offs04Maintenance check05Use cases06Alternatives
google /
elephant-goldfish-model
↗

Google plugin where a context-rich agent drafts the design spec and fresh zero-context subagents attack it

3.1kApache-2.0
Last commit1d ago
StatusActive
CategoryDev Tooling
Links
Documentation↗Discussion↗Issues↗Releases↗
Cloud Infrastructure
Ad

Deploy on DigitalOcean

Host your MCP servers, Docker containers, and AI agents with $200 free credit over 60 days.

Claim $200 Free Credit
© 2026 AgentIndex.app|Built by a 10-year iOS Developer.
Best by use caseAboutContactAdvertisePrivacy PolicyTerms of UseQYSGitHubBuy me a coffee ☕

Browse by Category

MCP ServersCode AssistantWorkflow AutomationRAG / Knowledge BaseMulti-AgentBrowser AutomationLLM InfraDev ToolingObservability

Not affiliated with Anthropic, OpenAI or Microsoft.