What is computer-use-mcp AGENTS.md?

Repository instructions for computer-use-mcp, a project that gives agents controlled access to macOS computer actions. They define safety boundaries, command rules, ownership areas, and verification expectations.

Use them when changing or testing computer-use-mcp code, especially when deciding whether a command affects the live macOS desktop, accessibility permissions, screen recording, clipboard, or input delivery.

They help prevent tests or demonstrations from controlling real applications, the cursor, the clipboard, or windows without approval. They also encourage non-mutating checks such as building, testing, and checking command health.

How minghinmatthewlam/computer-use-mcp configures Codex and OpenCode

Reuse

Borrowing it

Nothing to install: this file belongs to minghinmatthewlam/computer-use-mcp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/minghinmatthewlam/computer-use-mcp/main/AGENTS.md

Clone the repo

git clone --depth 1 https://github.com/minghinmatthewlam/computer-use-mcp

Made for: Codex, OpenCode.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/minghinmatthewlam/computer-use-mcp/agents-md/github.svg)](https://agentmods.dev/instructions/minghinmatthewlam/computer-use-mcp/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/minghinmatthewlam/computer-use-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/minghinmatthewlam/computer-use-mcp/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.

Your own site · 80×15

<a href="https://agentmods.dev/instructions/minghinmatthewlam/computer-use-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/minghinmatthewlam/computer-use-mcp/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 978 This file is loaded in full into every session.

When invoked 978 The same file — it is already loaded in full.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00978	$0.00978
Opus 5.5	$0.00391	$0.00391
Sonnet 5.5	$0.00196	$0.00196
Haiku 4.5	$0.00098	$0.00098

Measured 2mo ago against content hash 53062e78a36b, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

computer-use-mcp 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.

AGENTS.md · 79 lines

How it starts

The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Guidelines

This repo exposes live macOS computer-use capabilities. Treat runtime changes as safety-sensitive even when they look like ordinary Swift edits.

Safety Boundaries

Do not run live GUI, TCC, app-control, clipboard, or global-cursor demos unless the user explicitly approves that run.
Default to non-mutating verification: swift build, swift test, and CLI help/version/health_report --json smoke checks.
Treat scripts/e2e_demo.py, doctor --prompt, call, serve against real apps, and any test using Accessibility, Screen Recording, clipboard, window management, or input delivery as mutating/local-only.
Never disable safety gates (COMPUTER_USE_MCP_NO_SAFETY=1), app leases, or daemon isolation to make a test pass unless the task is specifically about those controls and the final result verifies the safe default path too.

Command Discipline

Run each shell command as a separate tool call. Do not chain with &&, ||, ;, or pipes unless that shell behavior is the thing being tested.
Prefer rg/rg --files for discovery.
Do not push, publish, rewrite history, delete user data, or kill unrelated processes without explicit permission.

Ownership Map

Entry / wiring (main.swift, Serve.swift, Dispatch.swift, Call.swift, Doctor.swift, HealthReport.swift, Version.swift): subcommand routing, MCP stdio shim, shared dispatch funnel (rate limit + logging), CLI call harness, TCC doctor / structured health report, and the binary version stamp.
Sources/computer-use-mcp/Core/: capture, AX, input, targeting, policy, and config. Runtime or safety changes here require focused unit coverage plus a local real-surface verification plan.
Sources/computer-use-mcp/Tools/: MCP tool handlers and catalog. Keep schemas stable and verify CLI/MCP smoke behavior when changing contracts.
Sources/computer-use-mcp/Daemon/: shared engine process, socket protocol, and app leases. Verify multi-session and daemon fail-fast assumptions before claiming concurrency safety; when touching leases or the socket path, confirm concurrent sessions cannot interleave inside one app and that mutating tools cannot bypass daemon arbitration through in-process fallback.
Sources/computer-use-mcp/Overlay/: visible agent cursor. Keep it optional for headless/CI paths.
Sources/computer-use-mcp/ToolKit/: schema/tool-spec plumbing. Prefer small, backwards-compatible changes.
Sources/ComputerUseFixture/: deterministic AppKit truth-suite fixture (honest / liar controls, web pane, dense table) used by the release battery — see docs/fixture-app.md.
Tests/ComputerUseMCPTests/: pure unit tests only. Do not add live GUI/TCC tests to the default suite.
scripts/: local demos and manual preflight helpers. Document required user consent and side effects.
.github/workflows/: hosted CI. Keep it deterministic and avoid live GUI/TCC requirements on hosted runners.

Read the full file on GitHub · 79 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.

2mo ago First seen · 79 lines · 978 tokens per session scan A 53062e78a36b

Subscribe to this mod's changes

computer-use-mcp AGENTS.md is an instructions file published in the GitHub repository minghinmatthewlam/computer-use-mcp (52 stars, last pushed 2mo ago), licensed MIT. It adds 978 tokens to every session, about $0.0039 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.

computer-use-mcp AGENTS.md next to 6 instructions from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
computer-use-mcp AGENTS.mdminghinmatthewlam/computer-use-mcp	Instructions file	52	2mo ago	MIT	A	978	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

computer-use-mcp: Instructions file for Codex