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 commands/oxbshw/watch-skill/setup-watch-skillgit clone --depth 1 https://github.com/oxbshw/watch-skillWrote 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/commands/oxbshw/watch-skill/setup-watch-skill)<a href="https://agentmods.dev/commands/oxbshw/watch-skill/setup-watch-skill"><img src="https://agentmods.dev/badge/commands/oxbshw/watch-skill/setup-watch-skill.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 | $0.00055 | $0.01007 |
| Opus 5 | $0.00028 | $0.00504 |
| Sonnet 5 | $0.00011 | $0.00201 |
| Haiku 4.5 | $0.00006 | $0.00101 |
Grade C, and why
setup-watch-skill scanned grade C 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 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.
Downloads and executes remote codehighSupply chain
curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.
Windows, or `curl -fsSL https://astral.sh/uv/install.sh | sh` on macOS/Linux), Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
Windows, or `curl -fsSL https://astral.sh/uv/install.sh | sh` on macOS/Linux), How it starts
The opening of the file, as written. The whole thing — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/setup-watch-skill
Set up Watch Skill end-to-end so /watch and the watch-skill MCP tools work
in this agent and every other agent on the machine. Do the steps in order and
stop at the first hard failure, reporting the fix the tool prints.
Step 1 — Ensure the engine is installed
Watch Skill's brains are the watch-skill Python CLI; this plugin is the thin
Claude Code surface. Check whether it's already on PATH:
watch-skill --version
If that fails, install it with uv (which also bootstraps its own Python —
no system Python needed). Prefer uv; fall back to pipx, then pip:
# preferred: isolated tool install from PyPI
uv tool install "watch-skill[standard]" \
|| pipx install "watch-skill[standard]" \
|| pip install --user "watch-skill[standard]"
standard is frames, retrieval, and MCP. Use watch-skill[all] instead to
add OCR, local Whisper, the REST surface, and the browser THE LOOP drives —
larger, and not needed to answer questions about a video. Whichever you pick,
doctor names anything missing and prints the command that adds it.
If uv itself is missing, install it first (winget install astral-sh.uv on
Windows, or curl -fsSL https://astral.sh/uv/install.sh | sh on macOS/Linux),
open a fresh shell so PATH updates, then re-run the install line above.
Re-run watch-skill --version and confirm it prints a version before going on.
Step 2 — Self-healing doctor
watch-skill doctor
doctor downloads ffmpeg / yt-dlp into a managed bin dir on first run. Exit 0
means ready. If it reports fixes, let it apply them and re-run once. Only
surface a check to the user if it still fails after that.
Step 3 — Register the MCP server in every agent
watch-skill setup --yes
This detects Claude Code, Claude Desktop, Cursor, Codex, Windsurf, and Gemini
CLI, backs up each config it touches, and writes the watch-skill MCP server
into all of them with a surgical merge (no existing keys are dropped). It
prints exactly which files it changed and where the backups are.
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 First seen · 101 lines · 55 tokens per session scan C 90af237c385f
setup-watch-skill is a command published in the GitHub repository oxbshw/watch-skill (330 stars, last pushed today), licensed MIT. It adds 55 tokens to every session and 1,007 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
fieldtheory
Explain and drive the Field Theory CLI — bookmark-sourced seeds, repo-aware possibility runs that score ideas onto a 2x2 grid, and the interconnected .md files they leave behind. Trigger when the user asks about ft, bookmarks, seeds, ft possible (or the older name ft ideas), grids, dots/nodes, or how to turn saved…
setup-video-vision
Interactive setup wizard for claude-video-vision — configure backend, whisper, frames, and verify dependencies.
pr
Create a pull request targeting the correct branch per project conventions.
add-agent
引导新增一个 Agent 适配器。用法 /add-agent.
describe
Write the searchable vocabulary this repository does not contain — agent-authored descriptions for undescribed declarations, the single biggest lever on retrieval quality.
search
Search this repository in plain language and report the declarations that answer it, each with its file, exact line range and the words it matched on.