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.
npx agentmods add instructions/lingyuanli/multigen/claude-mdgit clone --depth 1 https://github.com/lingyuanli/MultiGenWrote 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.
[](https://agentmods.dev/instructions/lingyuanli/multigen/claude-md)<a href="https://agentmods.dev/instructions/lingyuanli/multigen/claude-md"><img src="https://agentmods.dev/badge/instructions/lingyuanli/multigen/claude-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.01388 | $0.01388 |
| Opus 5 | $0.00694 | $0.00694 |
| Sonnet 5 | $0.00278 | $0.00278 |
| Haiku 4.5 | $0.00139 | $0.00139 |
Grade A, and why
MultiGen 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 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.
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.
How it starts
The opening of the file, as written. The whole thing — 127 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.
Overview
MultiGen is a general-purpose AI Agent system for fully private deployments. It uses a Planner + ReAct multi-agent architecture with A2A (Agent-to-Agent) and MCP (Model Context Protocol) tool connectivity, and executes operations inside an isolated Docker sandbox.
Architecture
Next.js UI (3000) → FastAPI API (8000) → PostgreSQL / Redis / Tencent COS
↓
Docker Sandbox (8080)
Ubuntu + Chrome + VNC
Agent execution flow:
AgentService(application layer) receives a chat message and dispatches it to anAgentTaskRunnervia a Redis Stream task queue.AgentTaskRunnerrunsPlannerReActFlow, which coordinates two agents:PlannerAgent— breaks the user request into sub-steps (JSON output, no tool calls)ReActAgent— iteratively executes each step using tools, then summarizes
- Events are streamed back to the frontend via SSE (
/api/sessions/{id}/chat).
Event types (discriminated union in app/domain/models/event.py): plan, title, step, message, tool, wait, error, done.
Tools available to agents (app/domain/services/tools/): file, shell, browser, search, message (ask_user), image_generation, volcano_image, volcano_video, video_concatenation, model_3d, virtual_anchor, qwen_tts, audio_mixing, mcp, a2a.
LLM abstraction: app/domain/external/llm.py defines a Protocol. The only implementation is OpenAILLM (app/infrastructure/external/llm/openai_llm.py), which supports any OpenAI-compatible endpoint (DeepSeek, Volcengine, SiliconFlow, etc.).
DeepSeek thinking models: Models whose names start with deepseek-v4 are treated as reasoning models. When tool calls are present in a turn, reasoning_content must be included in subsequent history; when there are no tool calls, it is passthrough-safe. The codebase handles both cases with a compatibility guard in BaseAgent.
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.
- 4d ago First seen · 127 lines · 1,388 tokens per session scan A 11c14d2c1000
MultiGen CLAUDE.md is an instructions file published in the GitHub repository lingyuanli/MultiGen (401 stars, last pushed 1mo ago), licensed MIT. It adds 1,388 tokens to every session, about $0.0069 per session on Opus 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 2026-08-30.
Other instructions, from other repositories
microsandbox AGENTS.md
AGENTS.md instructions for superradcompany/microsandbox, covering agents.md, scope, project map, design principles and backward compatibility review.
sandboxd AGENTS.md
Instructions for tastyeffectco/sandboxd, covering agents.md — operating sandboxd, what this is, prerequisites, install and core api.
docker-nixuser AGENTS.md
Instructions for grigio/docker-nixuser, covering development guide, ci publishing (critical), multi-platform ci builds, flake auto-update and docker image.
anvil AGENTS.md
AGENTS.md instructions for tools-for-agents/anvil, covering agents.md — anvil, setup, the rules this repo is built on, tests and ci.
aicontainer AGENTS.md
Instructions for stefanoginella/aicontainer, covering agents.md, repository layout, image tag pinning (read before editing the cli), version drift / update notifications and project identity and host-side validation.
aicontainer CLAUDE.md
Instructions for stefanoginella/aicontainer, a project described as: Sandboxed devcontainer for running Claude Code, Codex, and OpenCode in bypass / auto-approve mode.