iterative-converging-audit

Repeatedly audit a full surface until a sweep returns zero new findings.

75|18|Updated Feb 19, 2026
One-click install
npx skills add https://github.com/JKHeadley/instar --skill iterative-converging-audit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: iterative-converging-audit
Source: https://github.com/JKHeadley/instar/tree/main/skills/iterative-converging-audit
Command: npx skills add https://github.com/JKHeadley/instar --skill iterative-converging-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill prevents false confidence from one-pass reviews by forcing repeated audits until the target pattern yields no new findings.

Core Features & Use Cases

  • Frames the target pattern, search surface, classification rules, fix policy, and convergence criterion before the first sweep.
  • Runs an audit → fix → re-audit loop so every round checks the full surface, not just the files already touched.
  • Records a durable ledger of findings and decisions so accepted items are explicit, auditable, and not mistaken for misses.
  • Use it for security audits, code reviews, compliance checks, research sweeps, dependency hunts, or any exhaustive find-all task.

Quick Start

Use the iterative-converging-audit skill to sweep the repository for every instance of the target pattern, record the findings in a ledger, fix them, and keep re-auditing until a clean pass returns no new discoveries.

Frequently Asked Questions about iterative-converging-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I ensure my code review finds every instance of a security vulnerability?

An iterative converging audit ensures exhaustive code review by repeatedly sweeping the full search surface, fixing findings, and re-auditing until a fresh pass returns zero new discoveries.

What is the best way to perform an exhaustive compliance check across a repository?

Performing an exhaustive compliance check requires an iterative audit loop that frames classification rules upfront, records a durable findings ledger, and re-audits the entire surface until convergence is reached.

Why does a single pass security review miss patterns in my codebase?

A single pass security review creates false confidence by checking files only once, whereas an iterative audit continuously re-checks the full surface to capture patterns missed in initial sweeps.

How do I track accepted findings during a research sweep so they are not mistaken for misses?

During a research sweep, a durable findings ledger explicitly records accepted items and decisions, ensuring auditable tracking so previously reviewed items are not mistaken for missed discoveries.

When should I use a convergence criterion for an audit?

Use a convergence criterion for any exhaustive find-all task, such as dependency hunts or compliance checks, where a single pass is unreliable and the audit must only end when a full-surface re-audit yields zero new findings.