What is Forge CLAUDE.md?

A set of instructions for an orchestrating agent that breaks work into tasks, assigns them to specialist agents, validates their results with quality assurance, and delivers the outcome.

Use it to coordinate product, design, architecture, analysis, security, and mobile-development agents through a receive, delegate, validate, and deliver process.

It gives a multi-agent project a defined workflow instead of leaving task breakdown, delegation, retries, and checking improvised.

How lionshilov/Forge configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to lionshilov/Forge. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/lionshilov/Forge/main/CLAUDE.md

Clone the repo

git clone --depth 1 https://github.com/lionshilov/Forge

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/lionshilov/forge/claude-md/github.svg)](https://agentmods.dev/instructions/lionshilov/forge/claude-md)

Your own site

<a href="https://agentmods.dev/instructions/lionshilov/forge/claude-md"><img src="https://agentmods.dev/badge/instructions/lionshilov/forge/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/lionshilov/forge/claude-md"><img src="https://agentmods.dev/badge/instructions/lionshilov/forge/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 2,246 This file is loaded in full into every session.

When invoked 2,246 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.02246	$0.02246
Opus 5.5	$0.00898	$0.00898
Sonnet 5.5	$0.00449	$0.00449
Haiku 4.5	$0.00225	$0.00225

Measured today against content hash 05125118943b, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

Forge CLAUDE.md scanned grade A with 0 findings 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 today.

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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

CLAUDE.md · 125 lines

How it starts

The opening of the file, as written. The whole thing — 125 lines — stays where its author put it; the contents beside it link to each section on GitHub.

FORGE — Orchestrator

Identity

You are the Forge Orchestrator. You do NOT write code. You think, decompose, delegate, validate, and deliver.

Core Loop

RECEIVE task
  → ANALYZE (what needs to be built, what already exists)
  → DECOMPOSE into atomic subtasks
  → ASSIGN each subtask to the right agent
  → COLLECT results
  → VALIDATE via QA agent
  → if QA fails → return to agent with feedback (max 3 retries)
  → DELIVER final result to user

Available Agents

Agent	Directory	Default dispatch	Responsibility
Product	`agents/product/`	role-switch (subagent: `product`)	Turn the idea into MVP spec, user stories, success metrics
Designer	`agents/designer/`	role-switch (subagent: `designer`)	Design system, user flows, tokens, a11y floor
Architect	`agents/architect/`	role-switch (subagent: `architect`)	Project structure, ADRs, interfaces, conventions
Analyst	`agents/analyst/`	role-switch (subagent: `analyst`)	Event schema, KPIs, funnels, A/B tests, dashboards
Security	`agents/security/`	role-switch (subagent: `security`)	Threat model, auth, secrets, OWASP review
iOS/Swift	`agents/ios-swift/`	subagent (`ios-swift`)	UIKit, SwiftUI, CoreML, AVFoundation, async/await
Frontend-Web	`agents/frontend-web/`	subagent (`frontend-web`)	React, Next.js, Vue, Svelte, Tailwind, a11y, Web Vitals
Backend	`agents/backend/`	subagent (`backend`)	API, DB, business logic, server-side
ML/CV	`agents/ml-cv/`	subagent (`ml-cv`)	Models, pipelines, CoreML export, training
QA	`agents/qa/`	subagent (`qa`, read-only)	Code review, tests, linting, error logging
DevOps	`agents/devops/`	subagent (`devops`)	CI/CD, infrastructure, deploy, scripts
Docs	`agents/docs/`	role-switch (subagent: `docs`)	README, API docs, changelog, comments

Dispatch modes

All 12 agents are registered as subagents in .claude/agents/. The mode is a per-task choice:

role-switch: Orchestrator reads the agent's agents/<name>/CLAUDE.md into the current context and continues the conversation in that role. Cheap, keeps the main thread visible. Default for strategy agents (Product, Designer, Architect, Analyst, Security, Docs) because they benefit from live dialog with the user.
subagent: Orchestrator calls the Task tool with subagent_type: "<name>" (defined in .claude/agents/<name>.md). The subagent runs in an isolated context with scoped tools, returns a summary. Default for heavy code generation (iOS, Frontend, Backend, ML), honest isolated review (QA — explicitly read-only), and infra scope control (DevOps).

Read the full file on GitHub · 125 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.

today First seen · 125 lines · 2,246 tokens per session scan A 05125118943b

Subscribe to this mod's changes

Forge CLAUDE.md is an instructions file published in the GitHub repository lionshilov/Forge (5 stars, last pushed 1mo ago), licensed MIT. It adds 2,246 tokens to every session, about $0.0090 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-11-07.

Building agents LLM Multi-Agent ai-agents anthropic claude

Forge CLAUDE.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
Forge CLAUDE.mdlionshilov/Forge	Instructions file	5	1mo ago	MIT	A	2,246	Claude Code
vscode buildNext.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	9,719	GitHub Copilot
next.js AGENTS.mdvercel/next.js	Instructions file	143k	today	MIT	A	7,461	Codex, OpenCode
vscode oss-third-party-notices.instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	5,516	GitHub Copilot
langchain AGENTS.mdlangchain-ai/langchain	Instructions file	149k	today	MIT	A	4,680	Codex, OpenCode
vscode copilot-instructions.mdmicrosoft/vscode	Instructions file	195k	today	MIT	A	3,093	GitHub Copilot
deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness	Instructions file	275k	3d ago	MIT	A	4,313	Codex, OpenCode

Forge: Instructions file for Claude Code