Elephant-Goldfish Model

Elephant-Goldfish Model is an open-source plugin for spec-driven development using deep-context drafting agents and zero-context subagents.

Elephant-Goldfish Model is an open-source plugin for spec-driven development with AI coding agents. It uses deep-context agents called the Elephant to load system history, architecture, and user journeys to draft specifications. Stateless, isolated subagents called the Goldfish then read only the draft spec and source code to stress-test the design from conflicting perspectives before code is written. The plugin guides users through brainstorming ideas, drafting design documents, reviewing readiness, and checking for drift after implementation. It includes specific skills for brainstorming, drafting, resolving findings, bootstrapping documentation, reviewing designs, running review loops, and generating drift reports. Engineers use it to move from rough ideas to verified specifications.

Elephant-Goldfish Model at a glance

Best for: Software engineer
Pricing: Free
Released: September 2026
Category: AI coding tool
Website: egm.dev

Key features

Brainstorm ideas and write concept briefs
Draft living design documents section by section
Resolve review findings and agree on changes
Bootstrap documentation for existing codebases
Review designs with fresh agents across five stages
Run configured reviewer groups for specific review tasks
Compare checked-in documents with source code for drift

Use it for

Draft technical specifications using full repository context
Stress-test design documents with stateless review agents
Generate documentation for existing codebases
Compare implemented code against design specifications

Alternatives to Elephant-Goldfish Model

jevgrep — Reduces coding agent costs by providing targeted, relevant repository context.
TRAE — It features an autonomous coding agent that drives the entire development process.
Golem — Provides type-safe agent abstractions and zero-dependency Go implementation.
cc-sdd — Explicit boundaries and independent review let humans and agents work in parallel safely.
Atomic Agents — Build predictable, modular AI applications with consistent, reliable outputs.
Sandcastle — Orchestrates AI coding agents in isolated Docker, Podman, or Vercel sandboxes.
Context Engineering Kit — Improves agent reliability and code generation quality with minimal token footprint.
LLMSchema — Provides AI coding agents with accurate database schemas without context bloat.
Reika — Manages small local models with careful context discipline and loop breakers.
Adhdev — Orchestrates agents across machines with automated, test-gated git merging.
Context7 — Provides up-to-date documentation for AI agents and code editors.
Agents Best Practices — Provides a production-safe reference implementation for agent systems.

All Elephant-Goldfish Model alternatives