Skip to content
Agentmart
Submit a tool

finding-to-audit

Promote a review finding into a pre-commit audit rule that catches the diff-inspectable issue class (import direction, pub widening, debug artifacts, dropped…

Skill by ultimatile

IndexedMIT
Write a review

Chalk Labs

Need help deploying finding-to-audit?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/ultimatile-finding-to-audit

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Promote a review finding into a pre-commit audit rule that catches the diff-inspectable issue class (import direction, pub widening, debug artifacts, dropped FFI output). Companion to bug-to-contract.

Use cases: documents, security

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/finding-to-audit

What's inside

Skill name
finding-to-audit
Description
Promote a review finding into a pre-commit audit rule that catches the diff-inspectable issue class (import direction, pub widening, debug artifacts, dropped FFI output). Companion to bug-to-contract.
License
MIT
Instructions
92 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Finding-to-Audit

## Inputs

The fix-commit lane uses a **root** on `diff-root`'s consumer contract, halt included; the review-findings lane uses none.

| Input | What to collect |
| -- | -- |
| Review findings | Actionable findings that resulted in code changes — the reviewer's framing of the issue |
| Fix commits | Branch commits whose subject signals a fix: `git log <root-rev>..HEAD --oneline --grep="fix" -i`, then `git show <sha>` |

## Procedure

### 1. Collect signals

For each finding or fix, note what the issue was and what changed.

### 2. Classify

Would a structured inspection of the diff have caught it?

- **Yes** → continue here.
- **No, only runtime testing would** → switch to `bug-to-contract`.

Examples that fit this skill:

- Import direction violation (lower module imports from higher)
- `pub` exposed beyond what the architectural rule allows
- Missing standard trait impl on a public type (e.g. `Display`, `std::error::Error`, `Send`/`Sync` on a public error)
- New behavior added but no test reads its outputs
- Hardcoded constant that should be a parameter
- Debug artifact (`dbg!`, trace `println!`, commented-out code)
- FFI output channel silently dropped
- Public enum without `#[non_exhaustive]` where future variants are expected

### 3. Identify the host audit skill

| Concern | Host |
| -- | -- |
| Universal code quality | `quality-list` |
| Quality of text the agent executes as instructions | `authoritative-text-rules` |
| FFI safe-wrapper rules | `rust-ffi-rule` (or language-specific equivalent) |
| Other domain-specific audit | The matching skill |

If no host fits, scaffold a new audit skill rather than forcing the rule into an unrelated one.

### 4. Draft the rule

Use the host's existing format, read from a sibling item in that host. For a rule-set SSOT, an item is its own file under the SSOT's `items/` directory, added to the SSOT's Items index in the same change, and shaped:

```
# <Item name>

<rule description: what to check, in 1–2 sentences>

**Concern conditions:**

- <specific diff-inspectable signal>
- ...

**N/A:** <when the rule does not apply>
```

A host may require more, in the body and in the index entry alike — `quality-list` tags each index entry with the lane that routes it, and `authoritative-text-rules` items additionally carry a Trigger, a Sweep the auditor executes, and a tie-break against sibling items. The sibling entry and sibling body you read are the authority on which parts are required; an item whose index entry omits what its host's runners select on is reachable by no lane.

The rule must be **diff-inspectable** — verifiable against `git diff` output, file paths, or grep, without running the code. If the check requires execution, it belongs in `bug-to-contract`.

### 5. Land the rule

Promoting a finding to a rule is a proposal to be reviewed, not an edit to apply from wherever the finding surfaced.

**Default — file an issue** against the host skill's repository via `file-issue`, handing it:

- **Host (proposed)** — the host from step 3 (`quality-list`, `authoritative-text-rules`, `rust-ffi-rule`, or a new skill); a hypothesis the issue confirms or redirects.
- **Finding** — what the review caught and where the fix landed; link the work-repo PR / issue for provenance.
- **Issue class** — the generalizable diff-inspectable class, not the one token that slipped.
- **Proposed rule** — the step-4 draft (item name, concern conditions, N/A carve-out).

The rule then lands as a reviewed change in that repository. On recurrence of a class that already has an open proposal, comment on it instead of filing a duplicate.

**Escape hatch — edit the host skill directly** only when filing is inappropriate: a skill not under version control, or an explicit opt-in for the case at hand. The same failure-mode-plus-proposal content still travels with the change (e.g. in the PR body). Update any output tables in the host file.

### 6. Backfill check

If the same issue class has surfaced more than once historically, strengthen the rule (tighter conditions, additional concern signals) rather than relying on a single occurrence.

## Principles

- **General over specific.** The rule catches the bug class, not the reviewer's exact wording.
- **Extend before adding.** Strengthening an existing audit item is preferable to adding a new one unless the topic is orthogonal to all existing items.

From ultimatile/development-skills, MIT.

More from ultimatile/development-skills

Signals

Stars
0
Forks
0
Open issues
71
Last commit (this folder)
Oct 11, 2026 (20 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from ultimatile

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get finding-to-audit running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying finding-to-audit?

CallGet help
Cite this page

finding-to-audit skill. Agent Mart. https://agentmart.dev/skills/ultimatile-finding-to-audit (last updated 2026-10-24)