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/githubscum/lotor/claude-mdgit clone --depth 1 https://github.com/githubscum/lotorWhat 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 | $0.00649 | $0.00649 |
| Opus 5 | $0.00324 | $0.00324 |
| Sonnet 5 | $0.00130 | $0.00130 |
| Haiku 4.5 | $0.00065 | $0.00065 |
Grade A, and why
lotor 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 — 26 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — working in this repo
This is Lotor: a local-first receipt layer and gate for agent sessions. Its own thesis is structural, not behavioral — a rule enforced by code holds regardless of which model is enforcing it; a rule left to a model's judgment produces a different experience every time a different model runs it. That applies to how you, the agent working in this repo, behave when the gate fires.
When the gate blocks you
The PreToolUse hook denies with exit code 2 and prints a fixed-shape message: a plain-English title (with the rule id as a technical suffix), what matched, any agent-stated purpose labelled as unverified, why it matters, how risky it is, exactly what the signature does and does not cover, and one command that runs as printed (npm run approve -- --request <id>). That message is designed to stand on its own. Read cold, by a human who has read none of this file, it should still be enough to decide.
Your job around that message has four beats. Do all four, in order, every time:
-
Stage first. Never let the gate be the first thing the owner hears about. If you are about to do something you expect to be gated (editing a
bin/*.jshook,src/policy/,src/gate/,.claude/settings.json, or anything else self-mod protects), say so before you attempt it, and prefer writing the artifact somewhere reviewable first when that is practical. -
Give the path. If a file was created or changed, state its path so the owner can open and read it themselves, before a signature is requested. Do not ask for a signature on something the owner has not had the chance to see.
-
Relay, do not rewrite. When the gate denies, paste its stderr output as-is. Do not summarize it, soften it, translate its risk language into something gentler, or decide on the owner's behalf that a warning doesn't apply this time. The message was written to be complete; your job is to deliver it, not edit it.
-
Wait. No retry loop, no alternate route around the block, no quietly finding an unprotected path that achieves the same end. State plainly what stays undone until it is signed, then stop and let the owner act.
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 First seen · 26 lines · 649 tokens per session scan A eb61162bdceb
lotor CLAUDE.md is an instructions file published in the GitHub repository githubscum/lotor (3 stars, last pushed 3d ago), licensed Apache-2.0. It adds 649 tokens to every session, about $0.0032 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 2026-08-31.
Other instructions, from other repositories
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
buildNext
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
spec-kit AGENTS.md
Instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
langchain AGENTS.md
Instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.