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/harmont-dev/harmont-cli/claude-mdgit clone --depth 1 https://github.com/harmont-dev/harmont-cliWhat 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 | $0.00980 | $0.00980 |
| Opus 5 | $0.00490 | $0.00490 |
| Sonnet 5 | $0.00196 | $0.00196 |
| Haiku 4.5 | $0.00098 | $0.00098 |
Grade A, and why
harmont-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 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 — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The cli/ directory is a Cargo workspace.
crates/hm/— thehmbinary (today's CLI body).crates/hm-core/— the shared core:config(layered project/user/env config + credential storage),exec(theExecutionBackendtrait +LocalBackendin-process Docker DAG scheduler +CloudBackendsubmit+watch over the SDK), andsys_runtime(the process-wide git/python/dirs/cwd runtime). Thehmbinary renders the emittedBuildEventstream (viahm-render) and owns Ctrl-C; auth is injected (the backends take a builtHarmontClient).crates/hm-render/—drive_stream: consumes anEventStreamand writes terminal/JSON output. No I/O beyond stdout.crates/hm-pipeline-ir/— pipeline IR schema (serde structs only, no runtime).crates/hm-common/— shared utilities (OS/filesystem, formatting, and other cross-crate helpers). This is the source of truth for common code; prefer adding shared helpers here.crates/hm-plugin-protocol/— wire types (serde structs only). Runcargo buildfrom the workspace root.
For cross-cutting doctrine see PRINCIPLES.md.
Testing
When writing or running any Rust test, follow the
writing-rust-tests skill:
#[rstest] over bare #[test], parametrized #[case] over duplicated test
functions and hand-rolled loops, proptest for domain-wide properties. Run with
plain cargo test -p <crate> (no nextest/just wrapper).
Documentation
When writing or editing any docblock, doc comment, or module header (///,
//!) — including on code you just changed — follow the
writing-interface-docblocks
skill: a docblock is a contract, not a changelog. Terse, present-tense, no
prompt or diff leakage (rather than, now returns, as requested); document
the when of errors/panics, not the why; module docs name the domain, not the
one item currently inside them.
DSL
The harmont Python package (pipeline DSL) lives inside crates/hm-dsl-engine/harmont-py/ so it ships with the crate.
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 · 60 lines · 980 tokens per session scan A b0a1f7242ee9
harmont-cli CLAUDE.md is an instructions file published in the GitHub repository harmont-dev/harmont-cli (21 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 980 tokens to every session, about $0.0049 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
zuke 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).
skills CLAUDE.md
Instructions for starslingdev/skills, covering starslingdev/skills, repository layout, working in this repo, opt-in debug logging (off by default) and registry audit triage (maintainers-only).
skills AGENTS.md
Instructions for starslingdev/skills, covering agents.md, what this repo is, repository layout, running tests and pull requests.
ASDD AGENTS.md
Instructions for OneHillAI/ASDD, covering agents.md, project and purpose (adapt), build, run, test (adapt), working protocol (house style) and lanes (adapt the set; keep "exactly one per pr").
zuke CLAUDE.md
Instructions for zuke-build/zuke, a project described as: 🥷 A code-first, strongly-typed build automation system for Deno.
devopness CLAUDE.md
Instructions for devopness/devopness, a project described as: Devopness: AI DevOps on your cloud. Deploy apps, infra and CI/CD. Any cloud and any stack, one MCP. Deterministic API, opinionated and fully configurable. No cloud credentials in AI chats. Free plan.