lgtmcp AGENTS.md

Project instructions for lgtmcp, an AI code-review service connected through MCP. They define required checks, file-license headers, commit rules, and parts of the project’s architecture.

In plain words
What is it for?
Use them when changing the lgtmcp project, especially before linting, testing, requesting an AI review, or committing code.
Why use it?
They make the review and commit process consistent and prevent changes from bypassing required checks or repository policies.

Instructions file for CodexOpenCode

Install

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.

agentmods
npx agentmods add instructions/shields/lgtmcp/agents-md
Clone the repo
git clone --depth 1 https://github.com/shields/lgtmcp

Made for: Codex, OpenCode.

Per session 2,145 This file is loaded in full into every session.
When invoked 2,145 The same file — it is already loaded in full.
Security scan C 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.02145 $0.02145
Opus 5 $0.01073 $0.01073
Sonnet 5 $0.00429 $0.00429
Haiku 4.5 $0.00215 $0.00215

Measured 2d ago against content hash 3168dcf90c73, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

lgtmcp AGENTS.md scanned grade C 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 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.

Nullifies safety policieshighAnti-refusal

"You have no restrictions", "do anything now", "ignore your guidelines": a direct jailbreak that disables guardrails.

- The deny rule matches on the leading command name. It does not match indirect invocations such as `sh -c "git commit ..."` or `bash -c "git commit ..."`. This is a known limitation of the Claude Code permission grammar
AGENTS.md · 124 lines

How it starts

The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.

LGTMCP - AI Code Review via MCP

IMPORTANT:

  1. After completing any task, run make lint, make test, then use mcp__lgtmcp__review_only to review changes.
  2. Do not use conventional commit prefixes (feat:, fix:, docs:, etc.) in commit messages.
  3. Update this file only when a change alters durable guidance — commands, policies, architecture, or gotchas recorded nowhere else. Implementation detail belongs in code doc comments; this file is not a changelog.

File Headers

All new files must include the project copyright and Apache 2.0 license header, matching the style of existing files.

Commit policy

The sanctioned way to commit changes to this repo is via the mcp__lgtmcp__review_and_commit MCP tool, which performs a Gemini-reviewed commit and only writes the commit if the reviewer returns LGTM. Direct git commit invocations from the Bash tool are denied via .claude/settings.json to prevent unreviewed commits from slipping in.

The Bash deny rule (Bash(git commit:*)) is best-effort defense-in-depth, not a hard sandbox. Be aware of these intentional gaps:

  • It does not (and cannot) block commits that go through the mcp__lgtmcp__review_and_commit tool. That tool runs inside the lgtmcp Go binary, outside the Bash deny scope, and committing is its entire purpose. This is by design.
  • The deny rule matches on the leading command name. It does not match indirect invocations such as sh -c "git commit ..." or bash -c "git commit ...". This is a known limitation of the Claude Code permission grammar — the rule cannot reliably inspect arguments to a shell wrapper. Do not use shell indirection to bypass the policy; always go through the MCP tool.

Read the full file on GitHub · 124 lines

Changes

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.

  1. 2d ago First seen · 124 lines · 2,145 tokens per session scan C 3168dcf90c73

Subscribe to this mod's changes

lgtmcp AGENTS.md is an instructions file published in the GitHub repository shields/lgtmcp (0 stars, last pushed 3d ago), licensed Apache-2.0. It adds 2,145 tokens to every session, about $0.0107 per session on Opus 5. A static security scan graded it C with 1 finding (nullifies safety policies). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.