forge CLAUDE.md

forge CLAUDE.md is an instructions file for coding agents from zjio26/forge. It costs 1,216 tokens per session, scanned C, original, MIT.

Repository instructions for Forge, a Claude Code plugin made of agent definitions, a coordinating skill, and a plugin manifest.

In plain words
What is it for?
They guide the multi-agent workflow, wave-based task processing, integration testing, and checkpointing progress for recovery.
Why use it?
They explain how the coordinator divides development into planning, implementation, testing, fixing, and learning stages.

Instructions file

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/zjio26/forge/claude-md
Clone the repo
git clone --depth 1 https://github.com/zjio26/forge

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for forge CLAUDE.md

README.md
[![agentmods](https://agentmods.dev/badge/instructions/zjio26/forge/claude-md.svg)](https://agentmods.dev/instructions/zjio26/forge/claude-md)
Your own site
<a href="https://agentmods.dev/instructions/zjio26/forge/claude-md"><img src="https://agentmods.dev/badge/instructions/zjio26/forge/claude-md.svg" alt="Measured on agentmods" height="20"></a>
Per session 1,216 This file is loaded in full into every session.
When invoked 1,216 The same file — it is already loaded in full.
Security scan C 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01216 $0.01216
Opus 5 $0.00608 $0.00608
Sonnet 5 $0.00243 $0.00243
Haiku 4.5 $0.00122 $0.00122

Measured 4d ago against content hash c5931638544f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

forge CLAUDE.md scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

rm -rf ~/.claude/agents/planner.md ~/.claude/agents/dev.md ~/.claude/agents/test.md ~/.claude/agents/learner.md ~/.claude/skills/forge
CLAUDE.md · 101 lines

How it starts

The opening of the file, as written. The whole thing — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

Forge is a Claude Code multi-agent collaborative development workflow distributed as a Claude Code plugin. It's not a traditional codebase — it's a collection of agent definitions (markdown files), a skill orchestrator, and a plugin manifest.

Architecture

Coordinator pattern: skills/forge/SKILL.md is the central orchestrator (Coordinator) that dispatches four specialized agents via the Agent tool. The Coordinator never reads intermediate file content — it only tracks paths and status to keep its context lean.

Coordinator (SKILL.md)
  ├── Planner (agents/planner.md) — sonnet, decomposes requirements into subtasks
  ├── Dev     (agents/dev.md)     — sonnet, implements + writes unit tests + fixes bugs
  ├── Test    (agents/test.md)    — haiku, unit tests (must pass) + integration tests (may skip)
  └── Learner (agents/learner.md) — haiku, extracts lessons → knowledge base

Key flow: Planner → (Wave loop: Dev → Test → Fix) → Full Integration Test → Learner. State checkpointing (state.json) at every step enables crash recovery. The Coordinator resumes the same agent instances (via SendMessage + agentId) rather than creating new ones.

Wave-Based Processing

For large requirements with many subtasks, the workflow uses waves to prevent context overflow:

  • Planner groups subtasks into waves based on dependency order (waves.json)
  • Each wave gets a fresh Dev + Test agent pair
  • Wave-level testing: unit tests + local integration only (no cross-wave business flows)
  • Handoff: each Dev agent writes a handoff file (modified files, key interfaces, decisions) for the next wave
  • Full integration test runs after all waves, using the complete unit + business test combo
  • Small projects (single wave) run identically to the original flow — zero overhead

Test Layering

  • Unit tests: must pass (no external deps). Dev agent writes them; Test agent verifies.
  • Integration tests: best-effort. If skipped due to environmental constraints (network, hardware, etc.), it's informational only — does not cause overall FAIL.

Read the full file on GitHub · 101 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 101 lines · 1,216 tokens per session scan C c5931638544f

Subscribe to this mod's changes

forge CLAUDE.md is an instructions file published in the GitHub repository zjio26/forge (11 stars, last pushed 3mo ago), licensed MIT. It adds 1,216 tokens to every session, about $0.0061 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other instructions, from other repositories