Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/lossless-claude/lcmnpx agentmods add commands/lossless-claude/lcm/lcm-dogfoodWrote 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/commands/lossless-claude/lcm/lcm-dogfood)<a href="https://agentmods.dev/commands/lossless-claude/lcm/lcm-dogfood"><img src="https://agentmods.dev/badge/commands/lossless-claude/lcm/lcm-dogfood/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/commands/lossless-claude/lcm/lcm-dogfood"><img src="https://agentmods.dev/badge/commands/lossless-claude/lcm/lcm-dogfood.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 | $0.00035 | $0.01194 |
| Opus 5 | $0.00017 | $0.00597 |
| Sonnet 5 | $0.00007 | $0.00239 |
| Haiku 4.5 | $0.00003 | $0.00119 |
Grade B, and why
lcm-dogfood 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 11d 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.
Hooks are registered in `.claude-plugin/plugin.json`, NOT `~/.claude/settings.json`. Verify all 4: How it starts
The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
lcm Dogfood — Live Self-Test Suite
Comprehensive self-test for the lcm system in a live Claude Code session. Covers all 9 CLI commands, 4 hooks, 8 MCP tools, and resilience scenarios across 39 checks in 10 phases.
Arguments: all (default), health, import, compact, promote, sensitive, pipeline, hooks, mcp, resilience, debug
Procedure
Execute each phase in order (or just the requested phase). For each check:
- Run the command or verify the condition
- Record: ✅ PASS, ❌ FAIL, or ⚠️ SKIP (with reason)
- On FAIL: capture error, check daemon logs (
~/.lossless-claude/daemon.log), continue - Produce the Scorecard at the end
- Write failures to
.xgh/reviews/dogfood-YYYY-MM-DD.md
Routing: Use ctx_execute (context-mode sandbox) for commands producing large output. Use Bash for short-output commands. Use MCP tools directly for Phase 8.
Consult references/checks.md for detailed pass/fail criteria for each check.
Phase Overview
| # | Phase | Checks | What it tests |
|---|---|---|---|
| 1 | Health | 3 | Daemon status, doctor, version |
| 2 | Import | 3 | Transcript ingestion + idempotency |
| 3 | Compact | 3 | Summarization + idempotency |
| 4 | Promote | 2 | Insight extraction + stats consistency |
| 5 | Sensitive | 5 | Pattern list/test/add/remove cycle |
| 6 | Pipeline | 2 | Full curate + diagnose |
| 7 | Hooks | 6 | Wiring verification + live tests |
| 8 | MCP | 8 | All 7 MCP tools + store-retrieve roundtrip |
| 9 | Resilience | 3 | Kill/restart/graceful degradation |
| 10 | Debug | 4 | Logs, PWD, DB existence, integrity |
Key Commands
All CLI checks use node dist/bin/lcm.js <subcommand>. If lcm is on PATH, use that instead.
Hook Verification
Hooks are registered in .claude-plugin/plugin.json, NOT ~/.claude/settings.json. Verify all 4:
SessionStart→lcm restoreUserPromptSubmit→lcm user-promptPreCompact→lcm compact --hookSessionEnd→lcm session-end
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.
- 11d ago First seen · 102 lines · 35 tokens per session scan B 0877ef74d662
lcm-dogfood is a command published in the GitHub repository lossless-claude/lcm (26 stars, last pushed yesterday), licensed MIT. It adds 35 tokens to every session and 1,194 once invoked, about $0.0002 per session on Opus 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 2026-08-30.
Other commands, from other repositories
prototype
You are building a proof-of-concept for the current Grainulator sprint. Read CLAUDE.md for sprint context and claims.json for existing research claims.
verify
Run repository verification using the verification-loop skill.
qa-changes
This skill should be used when the user asks to "QA a pull request", "test PR changes", "verify a PR works", "functionally test changes", or when an automated workflow triggers QA validation of code changes. Provides a structured methodology for setting up the environment, exercising changed behavior, and reporting…
test-coverage
Analyze test coverage and identify the highest-value gaps to fill.
tdd
A command that follows test-driven development (TDD), a method where you write tests before the code they check. It moves through writing a failing test, adding the smallest implementation, and then improving the code.
check-dev
Type-check a Z specification with fuzz.