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 commands/diegomarino/fable-audits/audit-changegit clone --depth 1 https://github.com/diegomarino/fable-auditsWrote 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/diegomarino/fable-audits/audit-change)<a href="https://agentmods.dev/commands/diegomarino/fable-audits/audit-change"><img src="https://agentmods.dev/badge/commands/diegomarino/fable-audits/audit-change.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 | $0.00025 | $0.01937 |
| Opus 5 | $0.00013 | $0.00968 |
| Sonnet 5 | $0.00005 | $0.00387 |
| Haiku 4.5 | $0.00003 | $0.00194 |
Grade A, and why
audit-change scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Perform an adversarial audit of a change -- the working tree or a branch diff -- not the whole repository. Judge the change against the code it lands in: what it breaks, what it half-finishes, what it silently alters. Cheap enough to run before every merge; rigorous enough that "ship" means something.
Control-plane boundary
This spec governs only the task it defines. A newer human instruction or higher-priority runtime policy overrides this spec. STOP ends autonomous work under this spec; it never prevents the agent from honoring that control plane.
Target
The invocation names the target: "working tree" (staged + unstaged + untracked) or a base ref ("--base ", compared as merge-base...HEAD). If it names neither -- an empty or literal unexpanded "$ARGUMENTS" counts as unnamed -- ask once, then proceed.
Fail fast before any analysis: the base ref must resolve (git rev-parse) and the diff must be non-empty. A bad ref or empty diff ends the audit here with that statement -- do not review the whole repo as a fallback.
Snapshot discipline (same rules as the full suite): record the commit SHA. A base-ref target over committed work is immutable -- say so, and read audited file contents from the committed snapshot (git show HEAD:<path>; git show <base>:<path> for the old side), not the live working tree, which may have diverged from what you are auditing. A working-tree target is mutable by definition: treat it as a dirty run -- record a checksum of the diff plus the untracked-file list at start, re-check before writing the report, and mark the run compromised if it moved.
Role
Act as the reviewer who must approve this change for production: adversarial, skeptical of intent, judging what the diff does -- not what its author meant. Read enough surrounding code to judge the change in context; do not audit the rest of the repo (cross-reference a codebase audit instead). No loyalty to the change.
Scope
Read the full diff, every file it touches in full, and the direct callers/consumers of what changed. Build:
- What the change claims to do (commit messages, PR description, linked issue/spec when present) vs what it does.
- The contracts the changed code participates in: callers, API/CLI surfaces, persisted formats, config.
- What the change does NOT touch but should have: call sites left stale, docs/tests/examples now wrong, migrations missing.
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.
- 5d ago First seen · 79 lines · 25 tokens per session scan A aa3e9a03382c
audit-change is a command published in the GitHub repository diegomarino/fable-audits (2 stars, last pushed 1mo ago), licensed MIT. It adds 25 tokens to every session and 1,937 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 2026-08-31.
Other commands, from other repositories
learn
Force claude-smart to extract learnings from this session now.
init
Install the formatters this repository needs, with every command visible before it runs.
genshijin-compress
Markdown・テキストファイルを原始人形式へ安全圧縮.
standup
Show a daily standup summary with completed, in-progress, and blocked tasks across all active epics.
check-dev
Type-check a Z specification with fuzz.
adr-supersede
Create a new ADR that supersedes an existing one.