implement-audit

A one-time adversarial code review for recently changed code, looking for missed cases, inconsistencies, and breakages.

In plain words
What is it for?
It is for reviewing substantial code changes, reporting failed reviews, and identifying fixes that the original implementation missed.
Why use it?
It provides an independent check before work is considered complete and prevents the same changes from being reviewed repeatedly in one session.

Skill for Claude CodeCodex

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 skills/ashkaan/contextium/implement-audit
Any agent
npx skills add Ashkaan/contextium --skill implement-audit
Clone the repo
git clone --depth 1 https://github.com/Ashkaan/contextium

Made for: Claude Code, Codex.

Per session 71 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,741 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 $0.00071 $0.02741
Opus 5 $0.00036 $0.01371
Sonnet 5 $0.00014 $0.00548
Haiku 4.5 $0.00007 $0.00274

Measured yesterday against content hash 96bec57f014b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

implement-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 yesterday.

The scan reads SKILL.md. This mod also ships 2 executable files (scripts/audit-dedupe.sh, scripts/run-automated-checks.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

templates/claude/skills/implement-audit/SKILL.md · 206 lines

How it starts

The opening of the file, as written. The whole thing — 206 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/implement-audit — the loop's code reviewer

Review code just built in this conversation. Be adversarial — assume things were missed and find them. Do not confirm the work is good; find what is wrong.

This is the single code reviewer for the loop. /implement fires it after building; /close fires it as a backstop for work that never went through /implement. Both callers hit step 0 first, so it runs once per session no matter how many verbs ask for it.

Critical

  • Once per session, enforced by a file, not by judgment. Step 0 reads a marker keyed on the session id. Without it, /close would re-review a diff /implement already cleared — a full reviewer call and a fix loop spent re-deciding settled code. The marker also stores the first run's trailer, so the second caller re-emits it rather than re-earning it.
  • The reviewer should not be the author. code-review.sh runs the review on a different model when one is installed. When none is (exit 3), fall back to a fresh-context agent and say so — reduced independence reported as full independence is worse than no review, because it is trusted more.
  • A reviewer that could not run is a FAILED review. Exit 1 and 124 mean the review did not happen. Halt and report it; do not emit a passing trailer.
  • Only review work from this conversation. Do not flag unrelated uncommitted changes, review the whole git status, or pad findings to avoid an empty result.

Scope

Establish BASE_SHA — the start of this session's commits, typically HEAD~N for the N commits you made. Everything after it, committed or not, is in scope. Everything before it belongs to another session.

step-0-session-dedupe — have we already audited this session?

bash .claude/skills/implement-audit/scripts/audit-dedupe.sh status

done → print the stored trailer, say the audit already ran this session, and exit. fresh → continue.

step-1-automated-checks — the deterministic half

Read the full file on GitHub · 206 lines

Files

What ships with it

2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. yesterday First seen · 206 lines · 71 tokens per session scan A 96bec57f014b

Subscribe to this mod's changes

implement-audit is a skill published in the GitHub repository Ashkaan/contextium (32 stars, last pushed 4d ago), licensed MIT. It adds 71 tokens to every session and 2,741 once invoked, about $0.0004 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 2026-08-30.