Borrowing it
Nothing to install: this file belongs to nocoo/pew. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/nocoo/pew/main/.agents/skills/ponytail-audit/SKILL.mdgit clone --depth 1 https://github.com/nocoo/pewWrote 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/nocoo/pew/ponytail-audit)<a href="https://agentmods.dev/skills/nocoo/pew/ponytail-audit"><img src="https://agentmods.dev/badge/skills/nocoo/pew/ponytail-audit/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/skills/nocoo/pew/ponytail-audit"><img src="https://agentmods.dev/badge/skills/nocoo/pew/ponytail-audit.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 | $0.00020 | $0.00655 |
| Opus 5.5 | $0.00008 | $0.00262 |
| Sonnet 5.5 | $0.00004 | $0.00131 |
| Haiku 4.5 | $0.00002 | $0.00065 |
Grade A, and why
ponytail-audit 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 2d 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
Pew project override
In this repository, extend the global ponytail audit beyond complexity to accounting correctness, privacy and compatibility. Do not simplify away a correctness guard to reduce line count. This override changes only Pew's audit; it does not edit global skills.
Run sh scripts/ponytail-audit.sh from the repository root. The wrapper disables dotenv loading and Bun's transpiler disk cache. It prints one deterministic JSON object to stdout. Do not redirect it to a report file. The audit reads repository source files and runs synthetic SQL against an in-memory database; it never opens a source database, Pew state/queue, private log, credential file or remote endpoint. It does not run tests, sync, reset, upload, login, migration tools or cleanup commands. Do not follow symlinks or scan dependency, build, coverage or cache trees.
Read the accounting contract and inspect every stage below, including callers and consumers. A green mechanical audit is not a completed human review.
| Stage | Review obligations |
|---|---|
| Parser | Pi compaction vs assistant usage; Hermes main vs background/aux/ACP ledger; failed, cancelled and incomplete calls; source/provider model labels; only whitelist metadata leaves a source |
| Cursor | Partial lines, rollover, fork copies, reset and source loss; evidence committed before cursor; replay cannot change previously allocated times |
| Spool | Stable device/event/group identities; absolute replacement; source revision ordering; atomic outbox intent; retained evidence across reset; generic errors |
| Upload | Separate evidence endpoint; no SUM of snapshots; retry and offline convergence; old server rejection retains pending evidence; auth errors never expose payloads |
| Worker/schema | Additive migration before code activation; unchanged legacy UPSERT; immutable event ownership/time/provenance; device/user isolation; old snapshots cannot overwrite; account deletion |
| API/UI | Every usage read includes the combined view; original bucket dimensions and UTC storage; local display; no zero-checkpoint activity; approximate timing is visible |
| Test evidence | Inspect synthetic regression assertions for each layer; check current full-suite and G1 results separately. Suite presence and cached passes do not establish a fresh full run |
The JSON includes an input content digest, stage checks, regression-suite digests, findings and thresholds. Exit 0 means zero mechanical errors, 1 means at least one failed rule, 2 means the audit could not complete. Do not force findings to zero or lower a threshold. Full tests and independent reviews remain separate completion requirements.
For manual findings, use the same fields: id, severity, stage, path, message. Keep messages to the concrete defect, impact and smallest viable fix. Add complexity tags (delete, stdlib, native, yagni, shrink) only when supported by a real simplification. Use repository paths, safe counts and hashes; never quote source payloads, secrets, prompts, responses or personal messages. Reply in the pane only. Audit and review apply no fixes.
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.
- 2d ago First seen · 27 lines · 20 tokens per session scan A d34506cd355e
ponytail-audit is a skill published in the GitHub repository nocoo/pew (14 stars, last pushed 3d ago), licensed MIT. It adds 20 tokens to every session and 655 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-11-04.
ponytail-audit compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| ponytail-auditnocoo/pew | Skill | 14 | 3d ago | MIT | A | 20 | Claude Code, Codex |
| imagegenopenai/codex | Skill | 133k | yesterday | Apache-2.0 | A | 113 | Codex |
| playwright-climicrosoft/playwright | Skill | 99k | yesterday | Apache-2.0 | A | 19 | Claude Code |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | 4d ago | MIT | A | 200 | Claude Code, Codex |
| systematic-debuggingobra/superpowers | Skill | 313k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | today | MIT | A | 95 | Claude Code |
| next-partial-prefetching-adoptionvercel/next.js | Skill | 143k | today | MIT | A | 103 | Claude Code |