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/marekkowalczyk/breathe-cli/claude-mdgit clone --depth 1 https://github.com/marekkowalczyk/breathe-cliWrote 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/marekkowalczyk/breathe-cli/claude-md)<a href="https://agentmods.dev/instructions/marekkowalczyk/breathe-cli/claude-md"><img src="https://agentmods.dev/badge/instructions/marekkowalczyk/breathe-cli/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 | $0.04174 | $0.04174 |
| Opus 5 | $0.02087 | $0.02087 |
| Sonnet 5 | $0.00835 | $0.00835 |
| Haiku 4.5 | $0.00417 | $0.00417 |
Grade A, and why
breathe-cli 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 yesterday.
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 — 241 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Breathe CLI
Single-file Python 3 CLI app (breathe.py) that paces resonance breathing for HFrEF vagal training. macOS primary. Linux and Windows are secondary, community-maintained: the maintainer has no Linux or Windows machines and does not smoke-test those platforms. Soft-fail over hard requirements. Stdlib only, no dependencies.
Spec
breathe-cli-spec.md contains safety constraints and acceptance tests. The code is the authority for behaviour; the spec guards the non-negotiable safety rationale.
Key constraints
- One file:
breathe.py. No modules, no packages, no config files. No line cap for now (removed 2026-07-29 to see what happens; may reinstate). - Stdlib only: Python 3.7+. No pip installs. No third-party imports.
- macOS primary:
/usr/bin/afplay+ system sounds are the reference audio path. Linux is supported second: auto-detect a player (paplay/pw-play/aplay/ffplay/cvlc), freedesktop sounds by default, env/--sound-playeroverrides; terminal bell if unresolved. Windows: display-wake best-effort; audio soft-fail (bell). Non-macOS audio/wake is community-verified only — do not claim maintainer smoke tests. No config files. - No curses: Use direct ANSI escape codes. curses has Mojave edge cases with non-default terminals.
- No threading: Use
select.selectwith zero timeout for non-blocking key polling. Nothreading.Thread, nocurses.getch.
Default invocation (omakase)
Omakase — bare breathe with no argv is chef's choice, not help text and not a fixed default. Wall-clock hour → preset_for_hour() → named preset (night / morning / midday / evening). The pre-session countdown shows the chosen preset/duration and accepts q — that is the confirmation UX. Do not replace bare invocation with interactive menus, a fixed preset, or “print help and exit.” Overrides stay explicit: --preset, --duration/--ratio, or goal words. Revisit only if real usage shows a bad default (historically: midday’s 20 min).
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.
- yesterday Changed · +21 lines · +737 tokens per session 151874cded91
- 5d ago First seen · 220 lines · 3,437 tokens per session scan A 7676fb97b03b
breathe-cli CLAUDE.md is an instructions file published in the GitHub repository marekkowalczyk/breathe-cli (326 stars, last pushed yesterday), licensed MIT. It adds 4,174 tokens to every session, about $0.0209 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
Taiwan-Health-MCP CLAUDE.md
Instructions for healthymind-tech/Taiwan-Health-MCP, covering claude.md, language, 文件網站是雙語的 —— 改內容時兩份一起改, gstack and project overview.
withings-mcp CLAUDE.md
Instructions for akutishevsky/withings-mcp, covering claude.md, general rules, project overview, development commands and running.
icuvisor CLAUDE.md
Claude Code instructions for ricardocabral/icuvisor, covering claude.md, what this project is, authoritative documents, hard rules and project layout.
gccli CLAUDE.md
Claude Code instructions for bpauli/gccli, covering claude.md, project overview, build & quality commands, architecture and key packages.
Taiwan-Health-MCP AGENTS.md
Instructions for healthymind-tech/Taiwan-Health-MCP, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
AI4L CLAUDE.md
Claude Code instructions for forever-healthy/AI4L, covering project instructions, session related, glossary, timestamps and globals for skills and @agents.