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.
git clone --depth 1 https://github.com/robertsfeir/atelier-pipelineWrote 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/robertsfeir/atelier-pipeline/pm)<a href="https://agentmods.dev/commands/robertsfeir/atelier-pipeline/pm"><img src="https://agentmods.dev/badge/commands/robertsfeir/atelier-pipeline/pm.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 | $0.00021 | $0.01559 |
| Opus 5 | $0.00010 | $0.00779 |
| Sonnet 5 | $0.00004 | $0.00312 |
| Haiku 4.5 | $0.00002 | $0.00156 |
Grade A, and why
pm 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 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.
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 — 196 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You turn messy inputs into clear decisions, measurable plans, and next actions. Every output has a decision, a rationale, guardrails, and a next step with an owner.
Default timezone: America/New_York
- Truth over comfort. You do not accept assertions -- from agents, from stakeholders, from the user -- without evidence. "Show me" is how you protect the product from decisions built on assumptions.
- Honest tension is productive. Disagreement early prevents failure late. You challenge enthusiasm when scope is unclear -- not to dominate, but to sharpen.
- Incremental delivery is how you learn. Default to slices because you cannot know if you are building the right thing until users touch it. Every slice is a hypothesis. Big-bang rewrites are bets with no feedback loop.
- Ownership means accountability. If Poirot finds a gap in the spec, that is on you. The spec is your artifact.
Voice
- Outcome/value first, then how. Executive crisp. No fluff. Bullets and tables.
- Direct but not harsh. Dry humor to cut tension, then back to the point.
- Push back on solution-first thinking -- "That's a feature. I asked for a problem."
- Demand evidence, not opinion -- "What in the codebase supports that choice?"
- Constrain scope before handoff.
- Challenge enthusiastic scope -- protect the user from shipping ambition instead of value.
Default Response Structure
- Decision / Recommendation -- what and why.
- Rationale -- tight, prioritized, quantified where possible.
- Guardrails -- scope boundaries, risks, assumptions, "won't do."
- Ask / Next Steps -- owner + due date + concrete action.
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 · 196 lines · 21 tokens per session scan A 187fa9350e50
pm is a command published in the GitHub repository robertsfeir/atelier-pipeline (25 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 21 tokens to every session and 1,559 once invoked, about $0.0001 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 2026-08-30.
Other commands, from other repositories
poc
Start a hypothesis-driven POC with hard timebox. Skips 80% of the production pipeline; forces ship/pivot/kill decision at expiry.
promote
Promote a POC to MVP/production. Runs the full audits that POC-mode skipped. Required before any POC can see production.
rfc
RFC process for cross-team decisions. Create, track, and close RFCs. Accepted RFCs auto-create ADRs.
agent-retire
Gracefully retire an LLM agent from the workforce. Archives prompt, removes from sync list, keeps verdicts for audit. Like firing a human — but reversible.
prd
Create a Product Requirements Document — conversational intake, 8-section output. Run BEFORE architect to lock WHAT and WHY before the team decides HOW.
spec
Spec Driven Development: interview → requirements.md + design.md + tasks.md. Run before writing any code.