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 aristoteleo/PantheonOS --skill desktopgit clone --depth 1 https://github.com/aristoteleo/PantheonOSWrote 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/aristoteleo/pantheonos/desktop)<a href="https://agentmods.dev/skills/aristoteleo/pantheonos/desktop"><img src="https://agentmods.dev/badge/skills/aristoteleo/pantheonos/desktop/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/aristoteleo/pantheonos/desktop"><img src="https://agentmods.dev/badge/skills/aristoteleo/pantheonos/desktop.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00089 | $0.02073 |
| Opus 5.5 | $0.00036 | $0.00829 |
| Sonnet 5.5 | $0.00018 | $0.00415 |
| Haiku 4.5 | $0.00009 | $0.00207 |
Grade A, and why
Atrium Desktop — driving apps 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 1mo 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 — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The Atrium Desktop
The user works on a desktop of windows. Apps are installed packages; each claims file types, exposes actions, may run its own Python backend, and ships a skill documenting its state contract. You drive all of it with five tools — the SAME windows the user sees and clicks.
The tools
desktop_apps()
# → what is INSTALLED: app_id, name, description, opens, actions, backend,
# skill path. Use it to name an app explicitly or to see what opens a
# given file type — never guess an app_id.
desktop_windows()
# → every open window: window_id, app_id, title, path, actions, controllable
desktop_open(path="/abs/path/to/file") # like a double-click:
# routes by extension, runs the app's own open pipeline (conversion,
# backend prepare). Returns window_id. NEVER serve_local_data a file
# just to view it — desktop_open does everything.
desktop_open(app="viv", state={...}) # open on a state instead;
# each app's skill documents its state shape
desktop_read(window_id) # current state, skill-shaped
desktop_update(window_id, patch) # deep-merge a state patch
desktop_set(window_id, state) # REPLACE the state
desktop_call(window_id, action, args={}) # run a named action
desktop_open(path=..., window_id=...) # show another file in it
desktop_call(window_id, "$close") # close the window
desktop_screenshot(window_id) # see what it shows
app_call(app_id, method, args={}) # an app's backend method,
# in the app's own process. app_registry() lists live method signatures.
Windows the user opened are first-class: find them with
desktop_windows(), then read/update/call exactly as if you opened them.
The Browser (a shared, real Chromium)
browser_open(url) starts a real Chromium page in the sandbox and shows it
to the user as a Browser window. The page is SHARED: the user sees it live
and can click, type, and log in; you drive the same page with
browser_goto / browser_click / browser_type / browser_read /
browser_screenshot. When a site needs credentials, open it, ask the user
to sign in in the Browser window, then continue on the now-authenticated
page. The profile (cookies, sessions) persists in the sandbox. Use
browser_read (text) or browser_screenshot + observe_image (pixels) as
your eyes; prefer leaving pages open for the user over closing them.
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.
- 1mo ago First seen · 193 lines · 89 tokens per session scan A 85fee1d1c43f
Atrium Desktop — driving apps is a skill published in the GitHub repository aristoteleo/PantheonOS (491 stars, last pushed today), licensed BSD-2-Clause. It adds 89 tokens to every session and 2,073 once invoked, about $0.0004 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
visualize
A Chinese-language guide for creating interactive charts, diagrams, simulations, visual explanations, dashboards, visual comparisons, SVG or Canvas artwork, and interface prototypes in a Nexus conversation.
data-analysis
Analyze datasets and create visualizations.
imaging-data-commons
Query and download public cancer imaging data from NCI Imaging Data Commons using idc-index. Use for accessing large-scale radiology (CT, MR, PET) and pathology datasets for AI training or research. No authentication required. Query by metadata, visualize in browser, check licenses.
molecular-cloning
Molecular cloning simulation and design. PCR amplicon prediction, restriction enzyme digestion, Golden Gate and Gibson assembly simulation, primer design, CRISPR sgRNA design, and plasmid annotation. For protein-level sequence analysis use biopython or esm; for database lookups use gene-database or ensembl-database.
flow-cytometry-analysis
Complete flow cytometry analysis pipeline. FCS file handling, compensation, manual/automated gating, immunophenotyping, CFSE proliferation analysis, cell cycle analysis (Dean-Jett-Fox), and apoptosis assays. Extends flowio with analytical workflows. For raw FCS parsing only use flowio.
datamol
Pythonic wrapper around RDKit with simplified interface and sensible defaults. Preferred for standard drug discovery including SMILES parsing, standardization, descriptors, fingerprints, clustering, 3D conformers, parallel processing. Returns native rdkit.Chem.Mol objects. For advanced control or custom parameters…