Borrowing it
Nothing to install: this file belongs to uHappyLogic/cairn. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/uHappyLogic/cairn/main/CLAUDE.mdgit clone --depth 1 https://github.com/uHappyLogic/cairnWrote 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/uhappylogic/cairn/claude-md)<a href="https://agentmods.dev/instructions/uhappylogic/cairn/claude-md"><img src="https://agentmods.dev/badge/instructions/uhappylogic/cairn/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.
<a href="https://agentmods.dev/instructions/uhappylogic/cairn/claude-md"><img src="https://agentmods.dev/badge/instructions/uhappylogic/cairn/claude-md.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.17123 | $0.17123 |
| Opus 5.5 | $0.06849 | $0.06849 |
| Sonnet 5.5 | $0.03425 | $0.03425 |
| Haiku 4.5 | $0.01712 | $0.01712 |
Grade B, and why
cairn CLAUDE.md scanned grade B with 1 finding 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 3d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
uv run pytest agy_tests # on-demand suite driving the installed agy against ~/.gemini/config/plugins/cairn; run before committing an Antigravity integration change How it starts
The opening of the file, as written. The whole thing — 136 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.
What this repo is
A plugin (cairn) for Claude Code, Antigravity, and Codex that brings a milestone's important questions up to be decided before work starts, so execution can then be handed to agents. It ships as a set of skills, a few agents, shared procedures, and the stdlib-only Python tools under tools/ (among them the open-question tool and the task-list tool, the only writers of each milestone's open-question document and of its two task files once they are created, and the fold tool through which the answer sweep records its decisions in requirements.md) that move a project from a vague idea through discussion, planning, completion, and archival — one milestone at a time.
Commands
uv run scripts/build_hosts.py # rebuild hosts/ from core/ after any change under core/ or scripts/hosts/
uv run scripts/build_hosts.py --check # drift gate: fails if the committed hosts/ trees differ from a fresh build
uv run pytest # Python tool suite under the pinned interpreter (.python-version)
uv run --no-project --python 3.9 --with pytest pytest # the same suite at the 3.9 floor; run both before committing a tool change
uv run pytest agy_tests # on-demand suite driving the installed agy against ~/.gemini/config/plugins/cairn; run before committing an Antigravity integration change
uv run scripts/set_version.py X.Y.Z # bump VERSION and its three mirrors; releases go through /release-plugin instead
Repository layout
Each entry names the file that documents itself; read that file rather than expecting the detail here.
AGENTS.md — symlink to CLAUDE.md; never edited directly
VERSION — bare MAJOR.MINOR.PATCH, the single source of truth; written by set_version.py, read by build_hosts.py
.claude-plugin/marketplace.json — the maintainer's directory marketplace pointing at ./hosts/claude; users install from the distribution repos
core/ — the host-neutral runtime layer (skills/, agents/, shared/, tools/); cross-references use {{PLUGIN_ROOT}}, no file names a host
core/skills/<name>/SKILL.md — one skill per directory
core/agents/*.md — the two dispatched subagents (complete-task, provide-alternatives-to-open-question)
core/shared/*.md — the shared procedures, each the single source of truth for one unit of work referenced by two or more runners
core/tools/open_questions.py — the stdlib-only (Python 3.9+) open-question tool; run `--help` for its subcommands and contract
core/tools/define_milestone.py — the stdlib-only (Python 3.9+) milestone-definition tool: numbers and slugs a new milestone, creates its directory with the four files (the empty open_questions.xml included), prints the commit subject and directory
core/tools/task_list.py — the stdlib-only (Python 3.9+) task-list tool, the only writer of TASKS_TODO.md and TASKS_DONE.md once created; run `--help` for its subcommands and contract
core/tools/requirements.py — the stdlib-only (Python 3.9+) fold tool: `fold` splices one decision under a named `###` subsection of requirements.md's `## Decisions` and `replace` rewrites the body of one existing subsection, called by the answer sweep only and not that file's sole writer; run `--help` for its subcommands and contract
scripts/hosts/<host>/ — one declarative host definition per host: settings.toml, a slots/ subdirectory holding one value file per wording slot, and that host's manifest, README, CONTRIBUTING pointer, and traffic-workflow templates
scripts/build_hosts.py — the one generic host build and validation gate (see its docstring and CONTRIBUTING.md ## Development)
scripts/set_version.py — writes VERSION and its three mirrors (marketplace.json, pyproject.toml, uv.lock) in lockstep; touches no git
tests/ — pytest suite of the tools under core/tools/, one flat test_<name>.py per subcommand or concern, golden fixtures under tests/fixtures/, plus test_build_rendering.py, the build's in-process rendering test, which runs under the pinned interpreter only, and the two fixed-call tests (test_sweep_compact_view.py, test_sweep_batched_call.py), which extract the answer sweep's two shell calls from the rendered Claude Code tree, each as the one fence holding a command only it carries (the batched call's conflict section likewise, from the one fence holding `replace`) with nothing added to the runtime file, and run them under bash, under both interpreters
agy_tests/ — on-demand suite driving the real agy CLI against the global Antigravity copy, outside testpaths and CI, seeded from tests/fixtures/agy_project/; see CONTRIBUTING.md ## Development
hosts/<host>/ — committed build output, never hand-edited; published verbatim into uHappyLogic/cairn-<host> at release
.claude/skills/release-plugin/ — maintainer-only release skill, outside core/ so no host renders it
milestones/ — this repo's own milestones (self-dogfooding); milestones/README.md holds the Current milestone: pointer
milestones/answer_decision_principles.md — project-wide principle store read by recommend-procedure.md, written only by capture-milestone-principle-updates
ideas/ — design drafts that precede any milestone, versioned so they can be revised in the open; holds the benchmark draft
.github/workflows/drift-gate.yml — CI: --check then both pytest runs in one job; actions pinned to SHAs bumped by hand
.github/workflows/traffic-badges.yml — daily traffic-badge workflow; its copies under scripts/hosts/ are kept in sync by hand (grep github-traffic-badge@)
docs/ — workflow.md (phase narrative, how skills commit, learning loop), skill-reference.md (one entry per skill), design-claims.md, ways-of-using-cairn.md (headless chains)
README.md, CONTRIBUTING.md, SECURITY.md, CODE_OF_CONDUCT.md, CHANGELOG.md, LICENSE — landing page, contributor guide (## Development is the build loop), root-only policy files, release notes written by /release-plugin
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.
- 3d ago Changed · +3 lines · +2,609 tokens per session 1cbc79ccc9f4
- 10d ago Changed · +4 lines · +1,818 tokens per session scan A → B b46ba56ffb27
- 14d ago Changed · +2 lines · +1,642 tokens per session dc5e9e028055
- 15d ago Changed · +1,634 tokens per session c77bcc40d274
- 20d ago Changed · +1,593 tokens per session 033e951ba9fb
- 24d ago Changed · +592 tokens per session 6486da183efd
- 27d ago Changed · +1 lines · +297 tokens per session 0f4067cefbfc
- 28d ago Changed · +1 lines · +556 tokens per session 1e79ac1fa3ea
- 29d ago Changed · +2 lines · +626 tokens per session 92914fb1e36b
- 1mo ago First seen · 123 lines · 5,756 tokens per session scan A ce6c74c521bf
cairn CLAUDE.md is an instructions file published in the GitHub repository uHappyLogic/cairn (3 stars, last pushed yesterday), licensed MIT. It adds 17,123 tokens to every session, about $0.0685 per session on Opus 5.5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deno copilot-instructions.md
Copilot instructions for denoland/deno, covering deno development guide for github copilot, network access, high level overview, key directories and key files to understand first.