corpus

A command that audits skills written by other people and measures how often the project’s lint rules flag them. It reports scores, averages, rule frequencies, examples, and whether each flagged case is a real defect or a false positive.

In plain words
What is it for?
Use it to evaluate lint rules against a collection of external skills before changing those rules.
Why use it?
It tests whether lint rules work on unfamiliar real-world material instead of only on hand-written examples. This helps find rules that are too broad without weakening checks merely to improve the score.

Command for Claude Code

Install

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.

agentmods
npx agentmods add commands/alinotfoundbtw/sounding/corpus
Clone the repo
git clone --depth 1 https://github.com/alinotfoundbtw/sounding

Made for: Claude Code.

Per session 9 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 226 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00009 $0.00226
Opus 5 $0.00005 $0.00113
Sonnet 5 $0.00002 $0.00045
Haiku 4.5 $0.00001 $0.00023

Measured 2d ago against content hash 2d85be2f00e9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

corpus 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.

.claude/commands/corpus.md · 25 lines

What it actually says

Audit every skill under $ARGUMENTS — work we did not write — and report.

For each skill: score and the rule codes that fired. Then aggregate:

  • mean and median score
  • how many scored 100
  • a count per rule code, worst first
  • for any rule firing on more than ~15% of the corpus, quote two real examples

Then judge each firing rule honestly: is it a real defect in their work, or a defect in our rule? Assume ours until proven otherwise. The first corpus run found a 46% false-positive rate and four defects, including a rule that flagged security guidance because it quoted an attack string.

Do not change any rule yet. Report first, with a recommendation per rule.

If a rule needs loosening, also state how you will verify it still fires on genuinely bad input — tuning until the corpus is green is the same failure wearing a different mask.

Changes

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.

  1. 2d ago First seen · 25 lines · 9 tokens per session scan A 2d85be2f00e9

Subscribe to this mod's changes

corpus is a command published in the GitHub repository alinotfoundbtw/sounding (4 stars, last pushed 7d ago), licensed MIT. It adds 9 tokens to every session and 226 once invoked, about $0.0000 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.