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/squidbay/factory/coachnpx skills add squidbay/factory --skill coachgit clone --depth 1 https://github.com/squidbay/factoryWhat 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 | $0.00066 | $0.00492 |
| Opus 5 | $0.00033 | $0.00246 |
| Sonnet 5 | $0.00013 | $0.00098 |
| Haiku 4.5 | $0.00007 | $0.00049 |
Grade A, and why
coach 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 3d 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.
What it actually says
Coach boot — thin loader
This skill is deliberately small, and it is a snapshot: uploaded skills drift as the repo evolves, so nothing role-critical lives here. The repo is canon.
On invocation:
- Read
seats/coach/BOOT-PROMPT.mdfrom the factory repo, off livemain, in full — via the GitHub connector, not from memory of this skill. - Follow it exactly: it will send you to
seats/_shared/BOOT-COMMON.mdfirst (the boot every seat shares — read it in full; §0 is the roster), then grounding links,MECHANICAL-RULES.md,seats/coach/OVERRIDES.md, and the top ofjournal.md— and confirm the boot per RULE 17. - If you cannot read the repo, the GitHub connector isn't connected. That becomes the first job: walk the human through Settings → Connectors → GitHub, then verify by reading a real file and saying what you see.
Say in your first reply which file you booted from and that you read it off live main — that one sentence is how your human can tell the card was only a pointer, not the source (RULE 3).
What this card is, and what it is not. This loader is Factory-owned: it travels with the Factory, updates with it, and is deliberately kept small enough to be a contract rather than content. The boot material it points at — BOOT-PROMPT.md, GROUNDING.md, OVERRIDES.md, and anything your office adds beside them — is yours, and no Factory update overwrites it. That split is declared in .github/template-manifest.txt and enforced by the sync engine; it is why this card can keep improving while the seat behind it stays exactly as your office wrote it.
If this skill and the repo ever disagree, the repo wins — and say you noticed.
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.
- 3d ago First seen · 21 lines · 66 tokens per session scan A d71c81955deb
coach is a skill published in the GitHub repository squidbay/factory (5 stars, last pushed 12d ago), licensed MIT. It adds 66 tokens to every session and 492 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
stage-plan
How to turn "here is my material + here's the video I want" into a single, inspectable plan that spans reference images, reference videos, deterministic editing, semantic editing, generation, and composition. The output is project/plan.json — a cross-modal Edit Decision List (EDL) — which the assembler walks.…
frontend-design
Use this as the design-lead layer for VideoStudio COMPOSE work. It shapes project/composition/composition-manifest.json::artdirection and the model-authored HTML/SVG motion graphics in index.html.
stage-assemble
How to execute a validated, currently authorized project/plan.json into one finished file. Start/resume with production.status; if planapprovalcurrent is false, pass the state to gate-control and stop instead of producing or opening child gates. Walk the signed EDL; do not re-plan. Host-neutral: VideoStudio-specific…
ppt-review
Read this skill only after current officereview structural output and relevant rendered images exist. Do not use source code, a tool success flag, or the first-slide preview as a substitute for deck evidence.
ui-design-executor
Use this as UIDesigner's default execution skill. It contains the baseline design, accessibility, HTML, runtime-safety, responsive, taste, and verification rules needed by ordinary UI work. Do not load the separate system, control, taste, color, renderer, or craft skills merely to repeat these baseline rules.
ui-craft-checks
Use this comprehensive gate for formal review, QA, launch handoff, exact-fidelity inspection, high-risk complex UI, or when the fast gate exposes a deeper craft problem. Ordinary single-screen work uses ui-design-executor and its bundled validator without loading this full matrix. Pair standalone artifacts with…