Borrowing it
Nothing to install: this file belongs to ReviewStage/luke. 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/ReviewStage/luke/main/AGENTS.mdgit clone --depth 1 https://github.com/ReviewStage/lukeWrote 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/reviewstage/luke/agents-md)<a href="https://agentmods.dev/instructions/reviewstage/luke/agents-md"><img src="https://agentmods.dev/badge/instructions/reviewstage/luke/agents-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/reviewstage/luke/agents-md"><img src="https://agentmods.dev/badge/instructions/reviewstage/luke/agents-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 | $0.07116 | $0.07116 |
| Opus 5.5 | $0.02846 | $0.02846 |
| Sonnet 5.5 | $0.01423 | $0.01423 |
| Haiku 4.5 | $0.00712 | $0.00712 |
Grade A, and why
luke 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 7d 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 — 459 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent guide
Luke is a macOS Electron app, with a web app beside it, for working through a task with coding agents: the developer works a task through with them in typed conversations, and they write its plan down as they go and open its pull requests.
Commands
| Command | What it does |
|---|---|
./scripts/bootstrap.sh |
Install pinned workspace dependencies |
pnpm check:affected |
The portable checks for what changed since origin/main, replayed from cache where they already passed; optional before a push, since CI runs every check |
./scripts/check.sh |
Every portable repository, migration, lint, type, and test check: before a release, for a broad refactor, or on request |
./scripts/run.sh |
Launch against live sessions, replacing the running plain instance (--fixture smoke, --keep-running); --instance <name> runs a named one beside it, with its own profile, and --preview is one named for its Preview |
./scripts/evidence.sh |
Write the fixture PNG under artifacts/ |
pnpm release:macos |
Local signed, notarized, verified DMG, zip, and update manifest |
pnpm lint:fix |
Repository formatting and safe lint fixes |
CI runs the portable check on Linux alone, and no workflow runs the tests on a
Mac: the release job (release.yml's macos-15 job, run on a v* tag push or
a manual dispatch) only builds, signs, notarizes, and verifies the app. A green
PR says nothing about the Mac and there is no macOS check to wait for.
Layout
Application logic lives in packages/. An app holds what is specific to
its platform or deployment and nothing else: Electron's main and preload
processes, the window and page roots, the Vercel function routes under
apps/web/server/routes/, the deployment shell beside them, the agents' eve
apps, apps/dev, and build and release scripts. A React component, control, or
repeated UI shape must live in packages/ui whatever app mounts it, even one
app alone, with no exceptions, and a duplicate is folded into one; an app or
packages/panel must compose those components and never hand-builds a control
(a raw <button>, input, tab, or card with its own press handling) outside
@luke/ui: the design system is one rigid set that every agent must use. A hook or DOM-free view
model lives in packages/ui when more than one app can mount it, the panel every client mounts, the desktop's panel window and the
web's tab alike, in packages/panel, the hosted service's logic in
packages/service, its contract and clients in packages/features, and a
wire value in packages/wire. A
.ts module whose imports name no module of its app and no platform
library (electron, react-dom, a DOM API, better-auth, eve,
@vercel/*) is logic and belongs in a package unless it is shell: an app
entry in knip.json, a runtimeEdges row in
tools/anti-slop/effect-edges.json, or a deployment or test-support file.
Review places a module and a component alike under this rule.
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.
- 7d ago Changed · -1 lines · -28 tokens per session f243b7815faf
- 8d ago Changed · -4 tokens per session 7043a22a5352
- 9d ago Changed · -2 lines · -66 tokens per session e236a9bd79eb
- 10d ago Changed · +4 tokens per session 65411c392f1d
- 10d ago Changed · +2 lines · +30 tokens per session 3e688e83e76c
- 12d ago Changed · +43 lines · +570 tokens per session da551ec5fa9a
- 13d ago Changed df83e085a9f5
- 14d ago Changed · +32 lines · +579 tokens per session 83d6c4de78b3
- 15d ago Changed · -29 lines · -505 tokens per session a783ce6135ce
- 16d ago First seen · 414 lines · 6,536 tokens per session scan A d96374c76e92
luke AGENTS.md is an instructions file published in the GitHub repository ReviewStage/luke (39 stars, last pushed 5d ago), licensed Apache-2.0. It adds 7,116 tokens to every session, about $0.0285 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 2026-08-30.
luke AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| luke AGENTS.mdReviewStage/luke | Instructions file | 39 | 5d ago | Apache-2.0 | A | 7,116 | Codex, OpenCode |
| 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 |