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/max-rousseau/codeforge-mcp/claude-mdgit clone --depth 1 https://github.com/max-rousseau/codeforge-mcpWrote 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/max-rousseau/codeforge-mcp/claude-md)<a href="https://agentmods.dev/instructions/max-rousseau/codeforge-mcp/claude-md"><img src="https://agentmods.dev/badge/instructions/max-rousseau/codeforge-mcp/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 | $0.00716 | $0.00716 |
| Opus 5 | $0.00358 | $0.00358 |
| Sonnet 5 | $0.00143 | $0.00143 |
| Haiku 4.5 | $0.00072 | $0.00072 |
Grade A, and why
codeforge-mcp 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 5d 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 — 34 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CodeForge MCP — Agent Notes
Container Names
All three services use explicit container_name entries in docker-compose.yaml: codeforge-mcp-server, codeforge-proxy, codeforge-sandbox. The sandbox name is hardcoded in mcp-server/src/sandbox.ts and in the Docker socket proxy whitelist (mcp-server/docker-proxy.cjs) — both constants and the compose entry must stay in sync.
APIs Are Image-Baked
apis/ is COPYed into both the mcp-server and proxy images at build time — there is no runtime bind mount. Changes to apis/ on the host require docker compose up -d --build mcp-server proxy to take effect.
MCP Transport
The server uses Streamable HTTP with stateful sessions (not SSE, not stdio). Each client connection gets a fresh McpServer + StreamableHTTPServerTransport pair, tracked by session ID. The SDK (v1.29+) forbids reusing a stateless transport across requests — do not revert to a single shared transport. Claude Desktop connects via npx mcp-remote http://localhost:8080/mcp.
Proxy Host Resolution
In transparent mode, mitmproxy's flow.request.host returns the destination IP, not the hostname — so the substitution addon must use flow.request.pretty_host (which prefers SNI / Host header) to look up per-host credential maps. Reverting to flow.request.host silently breaks all credential substitution because the configured host strings (e.g. api.tavily.com) never match the IP keys.
Proxy Privilege Drop
The proxy entrypoint runs as root only long enough to: generate the mitmproxy CA cert, configure iptables NAT rules, and chown -R mitmproxy:mitmproxy /certs. Then it exec su -s /bin/bash mitmproxy -c "mitmdump ..." to run mitmproxy as the unprivileged mitmproxy user (created in proxy/Dockerfile). The chown is required because mitmproxy creates mitmproxy-ca.pem (CA private key) at 0600 via its umask_secret() context — the running mitmproxy needs read access to the key for on-the-fly leaf cert generation.
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.
- 5d ago First seen · 34 lines · 716 tokens per session scan A 65dde177adaa
codeforge-mcp CLAUDE.md is an instructions file published in the GitHub repository max-rousseau/codeforge-mcp (11 stars, last pushed 3mo ago), licensed BSD-3-Clause. It adds 716 tokens to every session, about $0.0036 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
codex-autoresearch AGENTS.md
Instructions for TheGreenCedar/codex-autoresearch, covering agents.md, purpose and scope, canonical sources, stable product boundaries and source and package boundaries.
mcp-erpnext AGENTS.md
AGENTS.md instructions for Casys-AI/mcp-erpnext, covering agents.md, project overview, project structure & module organization, build, test, and development commands and run all tests (also: deno task test).
zapcode CLAUDE.md
Instructions for TheUncharted/zapcode, covering claude.md, git, read agents.md first, codebase orientation and how to add a new language feature.
zuke AGENTS.md
AGENTS.md instructions for zuke-build/zuke, covering agents.md, using zuke — the api, without guessing, tech stack, typescript 7 / tsgo and coding guidelines (non-negotiable).
mcp-erpnext CLAUDE.md
Claude Code instructions for Casys-AI/mcp-erpnext: See AGENTS.md for all repository guidelines, architecture, conventions, and task recipes.
sprites-ex CLAUDE.md
Claude Code instructions for superfly/sprites-ex, covering sprites elixir sdk, quick reference, common commands, install dependencies and compile.