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/lestrrat-ai/claude-code-plugins/agents-mdgit clone --depth 1 https://github.com/lestrrat-ai/claude-code-pluginsWhat 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 | $0.04022 | $0.04022 |
| Opus 5 | $0.02011 | $0.02011 |
| Sonnet 5 | $0.00804 | $0.00804 |
| Haiku 4.5 | $0.00402 | $0.00402 |
Grade A, and why
claude-code-plugins AGENTS.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 — 222 lines — stays where its author put it; the contents beside it link to each section on GitHub.
claude-code-plugins
This repo authors the agent instructions: plugins/** defines the skills that may be driving the
current session. The skills actually running in a session are the installed copies, not the working
tree. The consequences below are all non-obvious:
This is a SINGLE-USER, advisory workflow — calibrate severity to that
The gauntlet campaign, its lease, its heartbeat / adoption / refresh machinery, and these plugin skills serve ONE user driving their own runs. Do NOT chase airtight consistency or distributed-systems-grade correctness in the heartbeat / lease / adoption / refresh paths. A momentary inconsistency the single user creates is one the single user notices and resolves — an ACCEPTED RESIDUAL, not a gating defect. (The user's stated policy.)
- The lease is advisory and LOOSE by design. Its heartbeat-proof precondition gates the FIRST
acquire — taking a run; an owner's
refresh— including refresh of its own STALE lease — is not required to re-present proof. Findings of the form "a stale/matching refresh revives a run without re-proof", "two agents could briefly both drive one run", or "heartbeat/adoption/refresh could be made more consistent/airtight" are accepted residuals, NOT defects — the tool never promised otherwise. - Calibrate finding severity to what the artifact IS. A same-machine, single-user skill helper is not
a web server facing millions of hostile requests. A guard being incomplete against an input only the one
user could hand-edit into their own git-ignored state (e.g. a
1e99999float in their ownlease.json) is not, by itself, a defect — the user shooting their own foot is the user's call. Document the boundary; do not build machinery to defend against the single user. - This does NOT lower the bar on REAL guarantees. Fail-closed on malformed input, refusing a
non-owner's destructive op, and accurate agent-consumed docs still gate. The line is: "a real
guarantee is broken" gates; "consistency could be more airtight for the one user" is an accepted
residual. When a review keeps generating true-but-that-kind-of finding, it is over-reaching on scope,
not finding defects — bound the intent (
repair-pass.md, REPAIR-INTENT) rather than fixing them.
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 · 222 lines · 4,022 tokens per session scan A 82883d9429f7
claude-code-plugins AGENTS.md is an instructions file published in the GitHub repository lestrrat-ai/claude-code-plugins (18 stars, last pushed 3d ago), licensed MIT. It adds 4,022 tokens to every session, about $0.0201 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
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.