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/akramlam/cairnos/claude-mdgit clone --depth 1 https://github.com/akramlam/cairnosWrote 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/akramlam/cairnos/claude-md)<a href="https://agentmods.dev/instructions/akramlam/cairnos/claude-md"><img src="https://agentmods.dev/badge/instructions/akramlam/cairnos/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 | $0.01804 | $0.01804 |
| Opus 5 | $0.00902 | $0.00902 |
| Sonnet 5 | $0.00361 | $0.00361 |
| Haiku 4.5 | $0.00180 | $0.00180 |
Grade A, and why
cairnos 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 — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CairnOS - working notes for Claude
CairnOS is a local-first AI productivity desktop app ("Turn chaos into action."). It is a pnpm-workspace monorepo where a small Node "local engine" (Hono) owns a SQLite database and exposes a REST-ish HTTP API plus SSE. A React + Vite + Tauri desktop UI talks to that engine over HTTP, and a separate local MCP stdio server shares the same database through the shared core package. Everything runs on the user's machine - no cloud dependency.
Architecture
- The local engine owns the SQLite file. A Tauri webview can't open a native SQLite file directly, so the Node engine (
apps/server) holds the connection viabetter-sqlite3+ Drizzle. The UI reads/writes only through the engine's HTTP API. @cairn/coreis the shared brain. Services and the rule-based classifier live in core and are consumed by both the engine (HTTP handlers) and the MCP server (tool implementations) - one set of business logic, two front doors.- The UI is a thin client. The React app is mostly a TanStack Query client over the engine's API; server state lives in the engine, UI-only state lives in Zustand.
- Timestamps and IDs are plain TEXT. Timestamps are ISO-8601 strings and IDs are UUID v4 strings, stored as TEXT in the DB, sent as-is over the API, and rendered as-is in the UI - zero serialization mapping anywhere.
- WAL mode enables multi-process access. The engine and the MCP server are separate OS processes hitting the same
cairn.db; SQLite WAL mode lets them read/write concurrently.
┌──────────────┐ HTTP / SSE ┌─────────────────┐
│ Desktop UI │ ───────────────────────▶ │ Local engine │
│ (TanStack Q) │ localhost:4319/api │ (Hono, apps/ │
└──────────────┘ │ server) │
│ owns cairn.db │
┌──────────────┐ @cairn/core (services + classifier) │ better-sqlite3 │
│ MCP server │ ───────────────────────────────────────▶ │ + Drizzle (WAL)│
│ (stdio, 13 │ same SQLite file └─────────────────┘
│ tools) │
└──────────────┘
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 · 87 lines · 1,804 tokens per session scan A 272a71e22fdb
cairnos CLAUDE.md is an instructions file published in the GitHub repository akramlam/cairnos (0 stars, last pushed 2mo ago), licensed MIT. It adds 1,804 tokens to every session, about $0.0090 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 2026-08-31.
Other instructions, from other repositories
grounded-knowledge-engine AGENTS.md
Instructions for dimosgit/grounded-knowledge-engine, covering agent guide, agent operating contract, authorship and attribution, two npm trees and commands — engine (repo root).
TORQCLAW CLAUDE.md
Instructions for pilotwaffle/TORQCLAW, covering claude.md — torqclaw, 1. repo context, instruction precedence, non-torq boundary and 2. governing harness.
erga-mcp AGENTS.md
AGENTS.md instructions for Adr1an04/erga-mcp, covering erga mcp contributor and agent guide, start here, product boundary, change routing and privacy and safety.
TORQCLAW AGENTS.md
Instructions for pilotwaffle/TORQCLAW, covering the job, hard boundaries and coordination with the main lane.
grounded-knowledge-engine copilot-instructions.md
Instructions for dimosgit/grounded-knowledge-engine: Follow AGENTS.md as the canonical repository contract.
grounded-knowledge-engine CLAUDE.md
Instructions for dimosgit/grounded-knowledge-engine: This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.