Borrowing it
Nothing to install: this file belongs to TheSmokeDev/taskchad-os. 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/TheSmokeDev/taskchad-os/master/.claude/skills/desktop-agent/SKILL.mdgit clone --depth 1 https://github.com/TheSmokeDev/taskchad-osWrote 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/thesmokedev/taskchad-os/desktop-agent)<a href="https://agentmods.dev/skills/thesmokedev/taskchad-os/desktop-agent"><img src="https://agentmods.dev/badge/skills/thesmokedev/taskchad-os/desktop-agent.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 5.1 | $0.00133 | $0.00663 |
| Opus 5 | $0.00067 | $0.00331 |
| Sonnet 5 | $0.00027 | $0.00133 |
| Haiku 4.5 | $0.00013 | $0.00066 |
Grade B, and why
desktop-agent scanned grade B with 2 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 8d 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.
Unrestricted tool accessmediumExcessive agency
A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.
# Run any command in a new visible terminal window Runs shell commandslowCapability
Expected in a hook, worth knowing in a rule or an instructions file.
- `subprocess.Popen` with `shell=True` for `start` commands — child terminals are independent How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Desktop Agent
Native daemon at ~/.claude/live-chat/desktop_agent.py that watches a command queue
and executes actions on the real Windows desktop. Runs as pythonw (no console window).
All commands go through desktop_cmd.py which auto-starts the agent if not running.
Commands
DCMD="python ~/.claude/live-chat/desktop_cmd.py"
# Open a live chat terminal
$DCMD open-chat --ctx thehomie
$DCMD open-chat --ctx deployment-a
$DCMD open-chat # all contexts
# Open URL in default browser
$DCMD open-url --url https://example.com
# Open file in default app
$DCMD open-file --path "~/some-file.pdf"
# Run any command in a new visible terminal window
$DCMD run --command "npm start" --title "Dev Server"
$DCMD run --command "ssh [email protected]" --title "SSH"
# Windows toast notification
$DCMD notify --title "Build Done" --message "deployment finished"
# Agent management
$DCMD status # check if running + PID
$DCMD ensure # start if not running
How It Works
desktop_cmd.pyappends a JSON command to~/.claude/live-chat/commands.jsonldesktop_agent.py(running natively viapythonw) polls the file every 500ms- Agent reads the command and executes it with full Windows desktop access
- Single-instance: PID lock with ctypes
OpenProcess(not brokenos.kill) - Auto-start: VBScript in Windows Startup folder, also
desktop_cmd.pystarts it on demand
Anti-Zombie Guarantees
- Single instance enforced via PID file + live process check (ctypes, not os.kill)
atexithandler cleans PID file on exit- Stale PID file from dead process gets overwritten (checks
STILL_ACTIVEexit code) subprocess.Popenwithshell=Trueforstartcommands — child terminals are independent- Agent log at
~/.claude/live-chat/state/desktop-agent.log
Auto-Start
VBScript at shell:startup runs pythonw desktop_agent.py on Windows login.
No console window. If the agent dies, next desktop_cmd.py call restarts it.
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.
- 8d ago First seen · 69 lines · 133 tokens per session scan B c59d088b682d
desktop-agent is a skill published in the GitHub repository TheSmokeDev/taskchad-os (23 stars, last pushed yesterday), licensed MIT. It adds 133 tokens to every session and 663 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it B with 2 findings (unrestricted tool access, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
graphify
Use when the user invokes /graphify or asks to build, update, or query a knowledge graph from a folder of files (code, docs, papers, notes, images). Triggers: 'map this codebase', 'graph my notes', 'what is connected across these files', 'community detection', 'GraphRAG', incremental graph update, Neo4j/SVG/GraphML…
insights
Use when the user types /weekly or /monthly, asks for a weekly or monthly journal review or retrospective, asks 'how was my week' or 'how was my month', or wants trends, anomalies, or patterns across recent journal entries. Also fires when regenerating an insight report for a past week or month. NOT for writing…
setup-brain
Set up or upgrade an AI-powered Obsidian vault. Interviews you, builds your vault structure (or works with what you already have), creates your CLAUDE.md memory file, installs tools, and gets you journaling — all in one conversation. Also has a repair/upgrade path for existing users.
second-brain-mapping
Use when the user wants to map or remap their second brain, refresh the vault's queryable metadata index, extract structured frontmatter from typed notes (books, meetings, people, journals, goals), fill wikilink gaps, or surface cross-type insights and patterns no single file shows. Triggers: /second-brain-mapping…
sunday-review
Use when the user asks for the weekly meta-review or Sunday wrap-up of their week and vault: says /sunday, /sunday-review, 'let's do the weekly review,' 'Sunday review,' 'review my week,' 'weekly retro,' 'week in review,' 'end-of-week review,' or wants journals, patterns, vault health, and stale decisions reviewed…
coach
Use when the user says /coach (today, week, month, profile, log), asks what to do today for training, asks for a workout, weekly training plan, or longevity plan, wants to log a finished workout or check lift progression (PR, plateau, deload), or a scheduled morning task fires to drop today's workout into the…