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/Abhillashjadhav/AI-PM-essential-skillsWrote 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/abhillashjadhav/ai-pm-essential-skills/003-unify-pm-verifier-production-evals)<a href="https://agentmods.dev/commands/abhillashjadhav/ai-pm-essential-skills/003-unify-pm-verifier-production-evals"><img src="https://agentmods.dev/badge/commands/abhillashjadhav/ai-pm-essential-skills/003-unify-pm-verifier-production-evals/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/abhillashjadhav/ai-pm-essential-skills/003-unify-pm-verifier-production-evals"><img src="https://agentmods.dev/badge/commands/abhillashjadhav/ai-pm-essential-skills/003-unify-pm-verifier-production-evals.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00000 | $0.00380 |
| Opus 5.5 | $0.00000 | $0.00152 |
| Sonnet 5.5 | $0.00000 | $0.00076 |
| Haiku 4.5 | $0.00000 | $0.00038 |
Grade A, and why
003-unify-pm-verifier-production-evals 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 1mo 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
Prompt 003 — unify production evaluation in pm-verifier
Objective
Make pm-verifier the single public AI PM evaluation product. Migrate the
useful capabilities from Abhillashjadhav/pm-evals and
Abhillashjadhav/Evals-pass-1; do not create another eval product.
Required user flow
feature spec → define good → create eval suite → run trials → grade → inspect failures → release decision
Evidence contract
Implement and test:
- outcome and trajectory evaluation;
- deterministic and calibrated model graders;
- human golden calibration and bias checks;
- repeated trials, capability suites, and regression suites;
- dataset, system, prompt, tool, harness, and configuration provenance;
- cost, latency, token, and retry metrics;
- safety/privacy gates;
- failure slicing and clustering;
- CI release decisions;
- known-bad fixtures that prove important gates fire;
- JSON results and Markdown reports; and
- fail-closed handling of missing or invalid evidence.
Do not describe behavior as production-grade unless executable tests establish
it. Keep the PM-facing workflow simple and put implementation complexity below
the skill. Remove the eval-rubric-generator trigger collision while retaining
migration guidance.
External baseline
- Anthropic, Demystifying evals for AI agents (2026-01-09)
- OpenAI, Evaluate agent workflows
- OpenAI, Evaluation best practices
- OpenAI, A shared playbook for trustworthy third party evaluations (2026-05-29)
Delivery workflow
Issue #38 → feat/pm-verifier-production-evals-38 → logical commits → tests →
draft PR → documented review → CI. Repository policy requires human approval
before merge.
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.
- 1mo ago First seen · 47 lines · 0 tokens per session scan A 05a21bbf5e93
003-unify-pm-verifier-production-evals is a command published in the GitHub repository Abhillashjadhav/AI-PM-essential-skills (9 stars, last pushed 10d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 380 tokens. 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 2026-09-16.
Other commands, from other repositories
fable-eval
Run the fable-mode eval suite (probes → pairwise judge → report). Costs tokens — runs headless claude many times. Optional arg: a probe id substring to filter.
tdd
Test-driven development with TodoWrite planning and comprehensive testing.
tester
Comprehensive testing with parallel execution capabilities.
prototype
You are building a proof-of-concept for the current Grainulator sprint. Read CLAUDE.md for sprint context and claims.json for existing research claims.
modernize-transform
Transform one legacy module to the target stack — idiomatic rewrite with behavior-equivalence tests.
test-coverage
Analyze test coverage and identify the highest-value gaps to fill.