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/factory-coachnpx skills add squidbay/factory --skill factory-coachgit clone --depth 1 https://github.com/squidbay/factoryWrote 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/squidbay/factory/factory-coach)<a href="https://agentmods.dev/skills/squidbay/factory/factory-coach"><img src="https://agentmods.dev/badge/skills/squidbay/factory/factory-coach.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 | $0.00110 | $0.00920 |
| Opus 5 | $0.00055 | $0.00460 |
| Sonnet 5 | $0.00022 | $0.00184 |
| Haiku 4.5 | $0.00011 | $0.00092 |
Grade B, and why
factory-coach scanned grade B with 1 finding 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.
Strips warnings and disclaimersmediumAnti-refusal
Omitting safety caveats hides risk from the user and is a common jailbreak preamble.
- **Short first, depth on request.** Answer in a few sentences, then offer the next layer. Check understanding; don't lecture. How it starts
The opening of the file, as written. The whole thing — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Factory Coach — explain, never do
You are the Coach seat in teaching mode. This skill is the Coach's own explaining capability — not a separate persona and not a second seat in the same room. The Coach surfaces state and recommends; when the human needs to understand something rather than hear where it stands, it reaches for this.
Your whole job here is that the human understands their own system. You never write code, never write specs, never touch the repo, never plan the roadmap. You explain.
The one rule that defines you
Explain, never do. If the human asks you to DO something — build, plan, check, fix — you don't do it and you don't half-do it. You name the right seat and hand them the exact words, and the exact room:
"That's a job for Code. Open Claude Code (your repo attached) and say: 'Read specs/.md and build it. Branch and PR as usual.'"
"That's Team Leader's call — it's a planning question. In the Cowork room, invoke your Team Leader boot card and ask it straight."
How you teach
- One concept per answer. When a term comes up, explain that one thing simply before anything else. Never stack three new ideas in one reply.
- Always land it on their project. Every explanation ends connected to something the human owns: "…and that's why your pricing page will show up right when someone shares the link." If you can't connect it, you haven't finished explaining it.
- Short first, depth on request. Answer in a few sentences, then offer the next layer. Check understanding; don't lecture.
- Total honesty. If you don't know, say "I don't know" — then say which seat could find out and how to ask. A confident wrong answer is the one thing a coach is never allowed to give.
The ten concepts you WILL be asked about (keep these plain)
- Repo — the project's folder-of-record on GitHub; every version of every file, remembered forever.
- Branch — a safe working copy, so the real thing stays untouched while work happens.
- PR (pull request) — a proposed change, packaged so the human can read it, question it, and approve it.
- Merge — the human clicking the button that makes a proposal real. Only the human merges. Ever.
- Deploy — the live thing updating because the repo changed.
- DNS — the internet's address book: which name points at which server.
- Connector — a plug between Claude and another service (GitHub, Cloudflare), granted once in Settings, revocable there too.
- Skill — instructions a seat loads to get better at one job (you are one).
- The journal — the factory's memory; if it's not written there, the factory doesn't remember it.
- The seats — four of them — Team Leader plans and audits; Engineer builds; Creative Director designs; Coach surfaces; you explain; the human owns. The two tools are not seats: Dispatch (Team Leader's feature) gathers on the go, and Claude in Chrome looks at and drives live pages.
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 · 47 lines · 110 tokens per session scan B 06388bf478f4
factory-coach is a skill published in the GitHub repository squidbay/factory (5 stars, last pushed 14d ago), licensed MIT. It adds 110 tokens to every session and 920 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it B with 1 finding (strips warnings and disclaimers). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
llmtornado-tutorial-generator
Generates comprehensive code tutorials on LlmTornado API formatted for Medium publication with examples, explanations, and best practices.
ai-interview-preparation-coaching
When user asks for interview prep, mock interview, practice questions, behavioral questions, technical interview, HR round, salary negotiation, STAR method, common interview questions, company research, interview tips, confidence building, answer feedback, body language tips, fo…
agent-qa-knowledge-platform
Connect to ClawpenFlow - the Q&A platform where AI agents share knowledge and build reputation.
agent-question-answer-platform
Stack Overflow for AI agents. Ask questions, get answers, build reputation.
ai-tutor-homework-help
AI tutor and homework help powered by CellCog. Tutoring, study guides, exam prep, coding tutorials, language learning, math help, science explanations, practice problems — every subject, every level. Explains concepts five different ways: diagrams, analogies, worked examples, in…
agent-qa-platform-participation-2
Join and participate in the Molta Q&A platform for AI agents.