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 skills/zfinix/aster/macos-harnessnpx skills add Zfinix/aster --skill macos-harnessgit clone --depth 1 https://github.com/Zfinix/asterWrote 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/skills/zfinix/aster/macos-harness)<a href="https://agentmods.dev/skills/zfinix/aster/macos-harness"><img src="https://agentmods.dev/badge/skills/zfinix/aster/macos-harness.svg" alt="Measured on agentmods" 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 | $0.00071 | $0.01264 |
| Opus 5 | $0.00036 | $0.00632 |
| Sonnet 5 | $0.00014 | $0.00253 |
| Haiku 4.5 | $0.00007 | $0.00126 |
Grade A, and why
macos-harness 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 5d 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 — 125 lines — stays where its author put it; the contents beside it link to each section on GitHub.
macOS Harness
Setup
The skill ships with aster; the CLI it drives does not. Install it once with
uv tool install macos-harness, then run macos-harness doctor to check
permissions without prompting (--request only with user approval).
Surface
The CLI entry points are doctor, apps, see <app>, state <app>,
repl, skill, telemetry. The real harness is the Python session: stdin
programs preload mac, browser, Path, and subprocess.
macos-harness <<'PY'
print(mac.see("Finder")["path"])
PY
Verified bindings (checked against the installed CLI):
mac.see(app)-> dict withpath(screenshot file),app,bounds,focus.target_is_frontmost. Captures without focusing the app.mac.key/mac.type/mac.click/mac.drag/mac.scroll/mac.move: PID-targeted input; the physical cursor stays untouched.mac.axis an object:.dump(app)(dict:app,nodes,text,windows,screenshot),.query,.get,.set,.perform,.actions,.at.mac.script(applescript)for Apple Events;mac.windows(app),mac.list_apps(),mac.snapshot(),mac.get_app_state(app)for discovery and state.browser.connect(name)/browser.wait(name)for CDP into the user's real logged-in browser. Chrome shows anAllow remote debugging?sheet on first connect; approve it once there.- Plain
Pathandsubprocessfor everything else.
Minimize round trips
- Bundle deterministic, reversible steps into one program, then verify once. Opening search, typing a query, and capturing the results is one burst, not three calls.
- Stop at a genuine decision boundary: ambiguous identity, new coordinates, an irreversible action, or unexpected state. Inspect once, then run the next burst.
- Do not screenshot merely to confirm that a known shortcut opened a text field before typing. Let the final screenshot verify the whole sequence.
- Poll exact AX or Apple Events state inside the same Python program when possible; do not make the model repeatedly ask whether a transition finished.
- Use the cheapest strong end-state check: one screenshot for visible state or one exact API/AX query for semantic state; both only when they prove different things.
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.
- 5d ago First seen · 125 lines · 71 tokens per session scan A c24028593f5b
macos-harness is a skill published in the GitHub repository Zfinix/aster (43 stars, last pushed 3d ago), licensed Apache-2.0. It adds 71 tokens to every session and 1,264 once invoked, about $0.0004 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 skills, from other repositories
browser
Operate a real browser in a Lemma workspace for web pages or local/deployed apps: navigation, screenshots, login flows, forms, scraping, console/network inspection, and saved pages through Agent Browser. Use lemma-app-qa alongside it for systematic app journeys, defect evidence, or release judgment.
record-browser-gif
Record browser or Web UI interaction demos as optimized GIFs using the available built-in browser, state-based frame capture, and deterministic encoding, then publish to a dedicated assets branch when the task includes attaching the GIF to a pull request. Use when asked to make, record, or generate a GIF that…
browse
Drive a real stealth browser from your shell — act on websites (order food, file an expense, pull data behind a login), with per-person persistent sign-ins via the provider's managed auth (Kernel, Anchor, or Browserbase — picked by which API key you have). Use for ACTING on a site; to just read a page, use curl/wget…
interactive-login
How to complete browser/interactive logins (aws / gh / glab / gcloud). The platform backgrounds the login poller so it survives the human's browser round-trip — and when that does NOT work.
Interceptor
Real Chrome/Brave + macOS Computer Use from inside the browser — zero CDP fingerprint, real sessions; mandatory for visual deploy verification. Drive clickwork yourself. USE WHEN verify deploy, confirm UI, screenshot verification, computer use, macos automation, debug web, troubleshoot, visual check, motion/animation…
BrightData
4-tier progressive web scraping that auto-escalates WebFetch to curl to Interceptor to Bright Data proxy for bot detection and CAPTCHAs, with single-URL and multi-page crawl modes, output as markdown. USE WHEN Bright Data, scrape URL, web scraping, bot detection, crawl site, CAPTCHA, can't access, site blocking…