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/kiliman/codemesh/claude-mdgit clone --depth 1 https://github.com/kiliman/codemeshWhat 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 | $0.01351 | $0.01351 |
| Opus 5 | $0.00675 | $0.00675 |
| Sonnet 5 | $0.00270 | $0.00270 |
| Haiku 4.5 | $0.00135 | $0.00135 |
Grade A, and why
codemesh CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories โ prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency โ measured 2d 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 โ 164 lines โ stays where its author put it; the contents beside it link to each section on GitHub.
CodeMesh MCP Server
CodeMesh is a "Code Mode" MCP server that executes TypeScript code against multiple MCP tools. Think of it as a universal adapter that lets agents write code to orchestrate ANY MCP server.
๐ฏ Core Concept
Instead of exposing individual MCP tools, CodeMesh provides:
discover-tools- See what's available (context-efficient overview)get-tool-apis- Get TypeScript APIs for specific toolsexecute-code- Execute TypeScript that calls multiple toolsadd-augmentation- Document unclear outputs (self-improving!)
๐๏ธ Project Structure
packages/
โโโ codemesh-server/ # Main CodeMesh MCP server
โโโ client/ # CLI for testing
โโโ example-server/ # Demo HTTP server
โโโ weather-server/ # Demo stdio server
โโโ geocode-server/ # Demo stdio server
๐ Key Features
Multi-Server Support
- โ HTTP, stdio, and websocket transports
- โ Connect to multiple servers simultaneously
- โ Execute code that calls tools from different servers
- โ Works with any MCP-compliant server
Context Efficiency
- โ Tiered discovery prevents context pollution
- โ Load only the tool APIs you need
- โ Single execute-code tool vs dozens of individual tools
Self-Improving Architecture ๐
- โ Auto-augmentation: Agents document unclear outputs
- โ
// EXPLORINGcomment: Triggers mandatory documentation workflow - โ Nuclear option: Returns ERROR until augmentation created
- โ Proven effective: Agent B one-shotted what Agent A struggled with!
- โ Knowledge compounds: Each exploration helps future agents
Security
- โ MCP SDK compliant env var handling
- โ
Environment variable substitution:
${VAR:-default} - โ Principle of least privilege
- โ Safe to commit configs (use env vars for secrets)
๐ Auto-Augmentation Workflow
The Problem: Agents trial-and-error unclear tool outputs instead of documenting them.
The Solution: Make exploration without documentation an ERROR!
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.
- 2d ago First seen ยท 164 lines ยท 1,351 tokens per session scan A c6e90fae8a78
codemesh CLAUDE.md is an instructions file published in the GitHub repository kiliman/codemesh (26 stars, last pushed 10mo ago), licensed MIT. It adds 1,351 tokens to every session, about $0.0068 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
ironcurtain CLAUDE.md
Instructions for provos/ironcurtain, covering claude.md, general workflow, git & worktrees, git workflow and platform considerations.
ironcurtain copilot-instructions.md
Instructions for provos/ironcurtain, covering ironcurtain copilot review instructions, platform support, security invariants, ai sdk v6 api (breaking changes) and code conventions.
ironcurtain security-boundary.instructions.md
Instructions for provos/ironcurtain, covering security boundary review rules, mandatory checks and structural invariant order.
ironcurtain sandbox.instructions.md
Instructions for provos/ironcurtain, covering sandbox review rules and mandatory checks.
ironcurtain types.instructions.md
Instructions for provos/ironcurtain, covering types review rules and mandatory checks.
ironcurtain pipeline.instructions.md
Instructions for provos/ironcurtain, covering pipeline review rules and mandatory checks.