Borrowing it
Nothing to install: this file belongs to Youhai020616/stealth-cli. 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/Youhai020616/stealth-cli/main/AGENTS.mdgit clone --depth 1 https://github.com/Youhai020616/stealth-cliWrote 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/youhai020616/stealth-cli/agents-md)<a href="https://agentmods.dev/instructions/youhai020616/stealth-cli/agents-md"><img src="https://agentmods.dev/badge/instructions/youhai020616/stealth-cli/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/youhai020616/stealth-cli/agents-md"><img src="https://agentmods.dev/badge/instructions/youhai020616/stealth-cli/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 | $0.03000 | $0.03000 |
| Opus 5.5 | $0.01200 | $0.01200 |
| Sonnet 5.5 | $0.00600 | $0.00600 |
| Haiku 4.5 | $0.00300 | $0.00300 |
Grade A, and why
stealth-cli 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 2mo 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 — 197 lines — stays where its author put it; the contents beside it link to each section on GitHub.
stealth-cli — Agent Development Guide
Core reminder
Reason from first principles, end to end. When the evidence does not support a conclusion, say "no basis found" instead of presenting an inference as a finding. Read the current file before editing it; keep changes scoped to the request; prefer an existing repository pattern over a new abstraction.
Product direction
Stealth Browser is a Camoufox cloud browser and hosted-agent platform, matching Browser Use's product capabilities: hand a task to the platform's agent, or create and control a cloud browser through the API and SDK. Browser execution is stealth-cli / SDK — the component this repository ships today, also usable on its own as a CLI, SDK, daemon, and MCP server over one engine, Camoufox. The Agent Core is Penguin Harness (selected, not integrated). Freepost, Travel Agent, and other applications integrate as external consumers of the generic API and SDK. Whether the platform's modules live in this repository or another is not decided; the browser component stays independent either way.
The goal, the interactions the product is judged on, the current implementation status, the
requirements, and the scope it declines are the root of the documentation: SPEC.md.
Read it before proposing a feature; it settles whether the feature belongs, and at which level.
Hard rules
- Camoufox through
playwright-core's Firefox driver is the only engine. Neverchromium.launch(), never theplaywrightbundle, never a second engine behind a flag. The product's one claim rests on the engine (decision). - Do not guess the
camoufox-jsAPI. It is a small library with little public documentation. Readsrc/utils/browser-factory.jsandsrc/browser.jsfor actual usage; the launch call is reproduced indocs/architecture.md. - Nothing the browser component does leaves a footprint in the page. No marker attributes,
no globals, no injected panels. Perception runs in Playwright's utility world; if a feature
needs to draw or remember something, it does so outside the page or not at all. This is a
stated boundary, not a claim that the operation as a whole is undetectable: clicks are
synthetic and
isTrustedis false, and documents say so where it applies. - Named state is single-writer and fails closed. Leases live in module-private
WeakMaps and never on a handle; writes are validated with the brandedownsStateLock; stale locks and cross-process.claim/.tmp/.rollbackartifacts are reported with their exact path and never auto-removed (decision). - Every surface tells the same truth. A daemon-mode operation has the same result shape,
error code, and state effect as its direct-mode twin, or the difference is documented at the
command. Errors cross a transport only through
serializeErrorForTransport(); raw causes, credential-bearing URLs, and cleanup internals stay non-enumerable. - The browser component's dependencies are exactly
camoufox-js,playwright-core,commander,chalk,ora, andvitestfor development. Notplaywright, not yargs/meow/oclif, not a second image or DOM library. Adding one is a decision written underdocs/decisions/first. Platform modules, when they exist, carry their own dependency rule. - ESM, plain JavaScript with JSDoc, Commander. Never
require(), no TypeScript. New commands go insrc/commands/<name>.jsexportingregister<Name>(program); new extractors insrc/extractors/<engine>.jsexportingname,canHandle(url),extractResults(page, max). - Every change leaves the living documents true. A change that alters a boundary, a contract,
or a default updates
docs/architecture.mdand the affected guideline in the same commit; one that settles a trade-off worth remembering adds a note underdocs/decisions/. A document left describing the old behaviour is part of the defect, not a follow-up. - English is the working language of living documents — code, comments, commit messages,
SPEC.md, this file,docs/architecture.md,docs/guidelines/,docs/decisions/,docs/postmortem/,tasks/. Dated records underdocs/research/anddocs/design/may be in the author's language; see the writing rules.
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.
- 2mo ago Changed · +59 lines · +913 tokens per session abe71ab700ee
- 2mo ago First seen · 138 lines · 2,087 tokens per session scan A 21d082435214
stealth-cli AGENTS.md is an instructions file published in the GitHub repository Youhai020616/stealth-cli (45 stars, last pushed 2mo ago), licensed MIT. It adds 3,000 tokens to every session, about $0.0120 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.
stealth-cli AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| stealth-cli AGENTS.mdYouhai020616/stealth-cli | Instructions file | 45 | 2mo ago | MIT | A | 3,000 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 9,719 | GitHub Copilot |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 5,516 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | today | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | today | MIT | A | 4,680 | Codex, OpenCode |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 3,093 | GitHub Copilot |
| vscode model-prompts.instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 2,022 | GitHub Copilot |