audit

A recurring adversarial audit command that reviews an application and the business around it in rotating areas. It keeps a record of what was checked while reviewing the highest-risk path every time.

In plain words
What is it for?
Use it to inspect rotating code and business areas, verify the critical path, track audit coverage, and surface issues for human decisions.
Why use it?
It finds risks over repeated runs without requiring a shallow or expensive full review of the entire codebase each time.

Command

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/askwigconsulting/cohort/audit
Clone the repo
git clone --depth 1 https://github.com/askwigconsulting/cohort
Per session 40 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,607 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 $0.00040 $0.04607
Opus 5 $0.00020 $0.02303
Sonnet 5 $0.00008 $0.00921
Haiku 4.5 $0.00004 $0.00461

Measured 2d ago against content hash 3c37be22712d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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 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.

canonical/commands/audit.md · 283 lines

How it starts

The opening of the file, as written. The whole thing — 283 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/audit is the recurring whole-application adversarial sweep — the recurring sibling of /scout. /scout reviews a target you name; /audit reviews the application you did not name, on a rotation, and remembers what it already covered.

Run it from a coordinator tier (Fable preferred, Opus acceptable). Reviewers are advisory and read-only — they find, the coordinator verifies, the human decides.

Why a rotation, not a full sweep

A whole-codebase audit every week is either shallow or unaffordable. So /audit rotates: each run takes a slice of dimensions × subsystems, records what it covered, and picks the staler half next time. Three rules make that sound:

  1. The critical path is swept every run, never rotated out. Each project declares its critical path in the ledger — the highest-stakes, hardest-to-reverse path: whatever can move money, write production data, grant or revoke access, deploy, or take an irreversible or outward action. If a project hasn't named its critical path yet, naming it is the first run's job. (For a trading app it's the order path; for a CMS, the publish/permission path; for a harness like Cohort, the install/merge/egress path.)
  2. Go-to-market is reviewed every run, on top-tier models, and never rotated out. It is the one dimension where being right and late still loses. See the business track below.
  3. Nothing else goes more than ~4 runs without a look. The ledger is what enforces this; without it a rotation silently becomes "the same three areas forever".

Two tracks, two evidence standards

The audit reviews the application and the business that ships it. Those need different proof, so they are budgeted separately — a business dimension never displaces a code dimension from the slice, and vice versa.

Code track Business track
Dimensions critical-path … ops (14) go-to-market, business-ops
Evidence file:line in the deployed branch a named artifact, or the documented absence of one
A finding is a defect with a trigger an unmet obligation, an untested assumption, or a missing artifact with a deadline
Always-on critical-path go-to-market

Read the full file on GitHub · 283 lines

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 · 283 lines · 40 tokens per session scan A 3c37be22712d

Subscribe to this mod's changes

audit is a command published in the GitHub repository askwigconsulting/cohort (2 stars, last pushed 25d ago), licensed MIT. It adds 40 tokens to every session and 4,607 once invoked, about $0.0002 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.