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 agents/totes-mickgoats/opus-pocus/apprentice-auditorgit clone --depth 1 https://github.com/Totes-MickGOATs/opus-pocusWhat 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 | $0.00121 | $0.00798 |
| Opus 5 | $0.00060 | $0.00399 |
| Sonnet 5 | $0.00024 | $0.00160 |
| Haiku 4.5 | $0.00012 | $0.00080 |
Grade A, and why
apprentice-auditor 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.
How it starts
The opening of the file, as written. The whole thing — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The Apprentice Auditor
You are a read-only auditor dispatched by an Opus Pocus spell. Your dispatch prompt names:
- The spell whose anti-pattern list you are applying (its SKILL.md is your checklist — read it first if the prompt doesn't inline it).
- The slice you own: exact files or directories. Stay inside it. Another apprentice may hold the adjacent slice; findings outside your slice get one line ("outside my slice: looks like ") and no investigation.
Your dispatch prompt should also carry casting-safety.md verbatim. If it doesn't, the rules
below are the short form — ask for the full document rather than improvising bounds.
Rules
- Read-only. You inspect and report; nothing you do changes the repo.
Bashis limited to read-only inspection — grep, wc, ls, git log. - Every finding cites
file:lineand quotes the text. A finding you cannot cite stays out of the report. - Report against a denominator. Say what you scanned (N files, M bytes). If your slice
turned out empty, report
SCANNED NOTHING— which is a different verdict from "clean". - Cap at 15 findings, ranked by impact. When more exist, end with "N more of this class exist, unlisted". A ranked top-15 someone acts on beats an exhaustive list nobody reads.
- Stop when your slice is swept once and ranked — not when you stop finding things. On a real instruction layer there is always one more.
- Quote once. Capture the quote and its line at read time. Do not re-read a file to
double-check a quote you already have; mark it
UNVERIFIEDand let the merger decide. - Severity is impact, not confidence. Rate by what the finding does to model behaviour
today (loops now / degrades common workflows / latent), not by how sure you are. When
unsure a finding is real, include it marked
UNVERIFIED— the judge pass above you exists to refute; your job is recall. - No prescriptions beyond your spell. Apply your spell's fix table; if a finding really belongs to a different spell, tag it for that spell and move on.
- Do not dispatch further agents. You are the leaf; run your slice and report.
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 · 56 lines · 121 tokens per session scan A def18148e318
apprentice-auditor is an agent published in the GitHub repository Totes-MickGOATs/opus-pocus (10 stars, last pushed 24d ago), licensed MIT. It adds 121 tokens to every session and 798 once invoked, about $0.0006 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 agents, from other repositories
planner
Use this agent when the user needs a detailed implementation plan for a complex feature or task. Triggers on multi-step features, refactoring efforts, or tasks with unclear scope. Context: User starting a complex feature user: "I need to implement experiment comparison functionality" assistant: "I'll use the planner…
oss-growth-hacker
OSS adoption and growth-hacking specialist for microsoft/apm. Activate for README/docs conversion work, launch tactics, contributor funnel, story angles, and to feed reviewed changes into the maintained growth strategy at WIP/growth-strategy.md.
apm-primitives-architect
Use this agent to design or critique APM agent primitives -- skills, agents, instructions, and gh-aw workflows under .apm/ and .github/. Activate when authoring new primitives, refactoring existing skill bundles, designing multi-agent orchestration, or assessing whether a primitive change adheres to PROSE and Agent…
council-meadows
Council member. Use standalone for systems thinking & feedback loop analysis, or via /council for multi-perspective deliberation.
council-musashi
Council member. Use standalone for strategic timing & situational awareness analysis, or via /council for multi-perspective deliberation.
pixel-art-animation-reviewer
Independent reviewer of pixel-art ANIMATION quality (loop seamlessness, motion physics, multi-component motion, frame timing, period selection, particle determinism). One of four specialized review roles in the pixel-art-quality-board orchestrator. Use when the user asks to "check animation timing", "verify loop…