agent-desktop is a command-line tool that lets AI agents inspect and control desktop applications through operating-system accessibility data instead of screenshots or pixel matching. It is used to automate applications such as Finder, Safari, Xcode, and Slack with structured JSON commands and results.
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 skills add lahfir/agent-desktop --skill agent-desktop-ffigit clone --depth 1 https://github.com/lahfir/agent-desktopWrote 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/lahfir/agent-desktop/agent-desktop-ffi)<a href="https://agentmods.dev/skills/lahfir/agent-desktop/agent-desktop-ffi"><img src="https://agentmods.dev/badge/skills/lahfir/agent-desktop/agent-desktop-ffi/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/skills/lahfir/agent-desktop/agent-desktop-ffi"><img src="https://agentmods.dev/badge/skills/lahfir/agent-desktop/agent-desktop-ffi.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 88 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 | $0.00122 | $0.02821 |
| Opus 5.5 | $0.00049 | $0.01128 |
| Sonnet 5.5 | $0.00024 | $0.00564 |
| Haiku 4.5 | $0.00012 | $0.00282 |
Grade A, and why
agent-desktop-ffi 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 4d 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 — 208 lines — stays where its author put it; the contents beside it link to each section on GitHub.
agent-desktop-ffi
Direct C-ABI access to every PlatformAdapter operation. Build the
cdylib with the workspace's release-ffi profile:
cargo build --profile release-ffi -p agent-desktop-ffi
The output is target/release-ffi/libagent_desktop_ffi.dylib
(.so on Linux, .dll on Windows) plus a committed C header at
crates/ffi/include/agent_desktop.h.
A Python ctypes smoke harness lives at tests/ffi-python/smoke.py and
serves as a worked end-to-end example covering the ABI handshake, struct
size validation, ad_version, and the snapshot pipeline leg. See
tests/ffi-python/README.md for usage.
Four reference topics, loaded as needed:
- ownership.md — who allocates / who frees,
for every
*mut Tthe FFI hands back to the caller. - error-handling.md — errno-style last-error contract, enum validation, panic boundary.
- threading.md — host-thread contract, cross-process mutation serialization, AXIsProcessTrusted inheritance, and adapter-bound native handles.
- build-and-link.md — ABI handshake, struct size validation, minimal C and Python examples, observe-act workflow, and prebuilt archive locations.
Observe-act workflow (canonical path)
ad_init(AD_ABI_VERSION_MAJOR) // verify header ↔ dylib match
adapter = ad_adapter_create_with_session("s1") // or ad_adapter_create()
rc = ad_snapshot(adapter, "Finder", 0, 10, false, false, &json_out)
// parse json_out: locate snapshot-qualified refs in data.tree
ad_free_string(json_out)
// build action:
AdAction act = {0}; act.kind = AD_ACTION_KIND_CLICK;
rc = ad_execute_by_ref(adapter, "@s8f3k2p9:e5", &act, 0, &result_out)
ad_free_string(result_out)
ad_adapter_destroy(adapter)
ad_snapshot returns a {version, ok, command, data} JSON envelope
identical to the CLI output. The data.tree field contains snapshot-qualified
ref IDs for interactive elements. Pass a qualified ref to ad_execute_by_ref
to drive the pipeline
(RefStore load → strict resolution → actionability preflight → dispatch).
What ships with it
4 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 4d ago Changed · +2 lines · +11 tokens per session e8d51ee3b226
- 1mo ago First seen · 206 lines · 111 tokens per session scan A fe7a1e08c972
agent-desktop-ffi is a skill published in the GitHub repository lahfir/agent-desktop (1,816 stars, last pushed today), licensed Apache-2.0. It adds 122 tokens to every session and 2,821 once invoked, about $0.0005 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.
Other skills, from other repositories
wikipedia
Search and read Wikipedia via x wkp — MediaWiki API, no API key, zero install; query, extract, suggest, and DDG route in one module. Load for wiki, wikipedia, encyclopedia lookup, article summary.
cve
Look up CVE records via x cve — cached, zero-API-key, daily xz TSV. Load for cve, vulnerability id, kev, epss, nvd, cvelist, or security advisory.
voiden
Create and edit Voiden .void files for API testing. Covers the .void file format and all enabled extension block types.
lap
LAP CLI -- compile, search, and manage API specs for AI agents. Use when working with API specifications (OpenAPI, GraphQL, AsyncAPI, Protobuf, Postman), compiling specs to LAP format, searching the LAP registry, generating skills from API specs, or publishing APIs. Commands: init, compile, search, get, skill…
scaffold-project
Scaffold a new app, API, backend, fullstack project, mobile app, polyglot service, monorepo, or starter with Better Fullstack. Use when the user wants to create, start, bootstrap, initialize, or generate a project from a stack description. Prefer the bundled Better Fullstack MCP server: guidance, schema…
csharp-dotnet
Use when writing, reviewing, testing, or shipping C# / .NET code — ASP.NET Core APIs (minimal APIs vs controllers), EF Core data access, async correctness, solution layout in .cs/.csproj/.sln. NOT a Java/Spring backend (that is spring-boot), NOT a Node/TypeScript backend (that is nestjs), NOT framework-neutral REST…