Borrowing it
Nothing to install: this file belongs to chankov/agent-fleet. 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/chankov/agent-fleet/main/CLAUDE.mdgit clone --depth 1 https://github.com/chankov/agent-fleetWrote 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/chankov/agent-fleet/claude-md)<a href="https://agentmods.dev/instructions/chankov/agent-fleet/claude-md"><img src="https://agentmods.dev/badge/instructions/chankov/agent-fleet/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/chankov/agent-fleet/claude-md"><img src="https://agentmods.dev/badge/instructions/chankov/agent-fleet/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 | $0.02784 | $0.02784 |
| Opus 5.5 | $0.01114 | $0.01114 |
| Sonnet 5.5 | $0.00557 | $0.00557 |
| Haiku 4.5 | $0.00278 | $0.00278 |
Grade A, and why
agent-fleet 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 1mo 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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
agent-fleet
This is the Agent Fleet project — a Pi-centered multi-agent orchestration system (agent-hub dispatcher, herdr fleet control plane, coms peer messaging, Hermes remote control) plus a library of production-grade engineering skills for AI coding agents. Lifecycle skills are partly vendored from upstream addyosmani/agent-skills — see docs/UPSTREAM-SKILLS.md.
Project Philosophy
Read docs/PHILOSOPHY.md before proposing or revising Agent Fleet plans, architecture, or process changes. Plans and design reviews must include a proportionate alignment note, distinguish current guarantees from proposals, and surface conflicts for a maintainer decision. Keep the principles in that canonical document rather than duplicating them here.
Project Structure
bin/ → npm CLI: cli.js (agent-fleet entrypoint) + the deterministic installer engine in lib/ — manifest.js (catalogue), state.js (record + hashing), verify.js (evaluation), plan.js (decision), apply.js (the ONLY module that writes to a workspace), merge-forms.js (justfile region + settings.json keys); plus doctor/detect-agent/personas, build-manifest.js, snapshot-version.js, test/ (node --test)
install-manifest.json → GENERATED catalogue of every installable artifact — never hand-edit; run `node bin/build-manifest.js`
manifest-meta.json → the hand-edited half: groups, profiles, recommendations, consent classes, companion/requires/pinnedBy wiring, operator steps
skills/ → Agent Fleet-native + customized skills (SKILL.md per directory); shadows same-named vendored skills
vendor/agent-skills-upstream/ → Pristine upstream skill import at a pinned SHA — NEVER edit in place; policy + update procedure in docs/UPSTREAM-SKILLS.md
agents/ → 15 reusable agent personas, canonical pi frontmatter; installed as verbatim copies into a workspace's `.pi/agents/personas/` (no per-agent translation — pi is the only install target)
.pi/agent-fleet/ → THE FLEET RUNTIME, mirrored verbatim into a workspace. Source layout == installed layout, which is what lets the installer copy it with `preserveLayout`. Sub-trees:
.pi/agent-fleet/scripts/ → Standalone scripts (team-up herdr launcher for reusable coms peers; peer-launch single-peer `just fleet peer` launcher; .pi/agent-fleet/scripts/lib/ pure fleet modules under node --test)
.pi/agent-fleet/scripts/workflows/ → Headless `just flow` graphs and their code-owned phase, envelope, gate, permission, quality, change-capture, trace, and scoped TypeScript runtime
.pi/agent-fleet/hermes/ → Hermes-facing skills (hub-conductor, hub-liaison) for remote conduction — see docs/coms-hermes-bridge.md
.pi/agent-fleet/hooks/ → Session lifecycle hooks (the Claude Code bridge Stop hook; registering it in `.claude/settings.json` stays the user's step)
(a workspace also gets .pi/agent-fleet/docs/workflows.md — the one doc that installs)
justfile → Recipes to launch pi with each harness plus the single public `flow` entry point
.changeset/ → Pending changesets; rolled into CHANGELOG.md + version bump by `changeset version`
.versions/ → Per-version artifact snapshots used by the version-aware update flow (snapshot-version.js)
.github/workflows/release.yml → On merge to main: opens "Version Packages" PR or runs `changeset publish`
.pi/prompts/ → pi-native lifecycle prompt templates
.pi/extensions/ → always-on pi utility extensions, auto-discovered by pi (mcp-bridge, chrome-devtools-mcp, compact-and-continue, btw, pi-voice-stt). pi-voice-stt is gated/optional — it binds its Alt+S hotkey only when an STT provider is configured, otherwise it is a no-op. The package update check is a Fleet Core helper (`.pi/harnesses/lib/update-check.ts`) scheduled by damage-control-continue, not an auto-discovered extension.
.pi/harnesses/ → selectable pi session harnesses — NOT auto-discovered; loaded explicitly via the justfile or `pi -e` (`just fleet hub` loads Fleet Core before agent-hub; every native child gets damage-control-continue)
.pi/agents/ → pi YAML configs (teams, chains, peers, dispatch-policy) used by the orchestration harnesses, plus `personas/` where the installer writes the 15 personas; dispatch-policy.yaml routes dispatch_agent calls to same-name coms peers (e.g. the `runner: claude-code` reviewers) with native fallback
.pi/skills/ → pi-runtime skills (e.g. bowser browser automation)
.pi/damage-control-rules.yaml → rule set for the damage-control harness
references/ → Supplementary checklists (testing, performance, security, accessibility, observability); each installs as a companion of the skills that cite it
docs/ → ARCHITECTURE.md (runtime layers + module map), UPSTREAM-SKILLS.md (vendoring policy — update it when touching vendor/ or shadowed skills), MIGRATION-agent-fleet.md (major-release migration matrix), setup guides, agent-fleet-setup.md (per-project overrides + ownership), npm-install.md (deterministic lifecycle reference), claude-code-coms-bridge.md and coms-hermes-bridge.md (bridge references), plus pi-extensions.md and pi-specs/ for the pi extensions
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.
- 1mo ago Changed · +47 tokens per session 68b58498ac8a
- 1mo ago Changed · +8 lines · +85 tokens per session 1d5bc316bba6
- 1mo ago Changed · +2 lines · +162 tokens per session 5aefd01a0328
- 1mo ago Changed · +28 tokens per session 91e1fbeb5683
- 2mo ago First seen · 79 lines · 2,462 tokens per session scan A 83d1848dec37
agent-fleet CLAUDE.md is an instructions file published in the GitHub repository chankov/agent-fleet (20 stars, last pushed 12d ago), licensed MIT. It adds 2,784 tokens to every session, about $0.0111 per session on Opus 5.5. 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 2026-09-03.
agent-fleet CLAUDE.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| agent-fleet CLAUDE.mdchankov/agent-fleet | Instructions file | 20 | 12d ago | MIT | A | 2,784 | Claude Code |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | yesterday | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |