ctx-audit

A set of checks for AI context files—files that give coding agents project information and instructions.

In plain words
What is it for?
It helps verify context-file claims against the codebase and find stale references after files, versions, or modules change.
Why use it?
It reduces errors caused by outdated versions, file paths, dependency names, symbols, or project counts in those files.

Cursor rule for Cursor

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 rules/knitli/toolshed/ctx-audit
Clone the repo
git clone --depth 1 https://github.com/knitli/toolshed

Made for: Cursor.

Per session 0 Nothing until a file matches its globs; then the whole rule loads.
When invoked 336 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 $0.00000 $0.00336
Opus 5 $0.00000 $0.00168
Sonnet 5 $0.00000 $0.00067
Haiku 4.5 $0.00000 $0.00034

Measured yesterday against content hash 11ef56925097, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

ctx-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.

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.

plugins/ctx/cross-client/.cursor/rules/ctx-audit.mdc · 33 lines

What it actually says

Context Hygiene Rule

When you edit or read any AI context file matched by this rule's globs, apply these checks:

Before trusting claims in a context file

Verify key factual claims against the actual codebase:

  • Version numbers: Check Cargo.toml, package.json, pyproject.toml
  • File paths: Confirm the referenced files/directories exist
  • Dependency names: Confirm they appear in the manifest
  • Symbol names: Search the codebase for referenced types, functions, modules
  • Counts: Verify "N crates", "N languages", etc. by actually counting

After editing code that context files reference

If you renamed files, changed versions, restructured modules, or modified the architecture, check these files for stale references:

  • CLAUDE.md and AGENTS.md (root-level agent memory)
  • .serena/memories/* (Serena tool memories)
  • .specify/memory/* (spec-kit memory)
  • GEMINI.md (if not a symlink)

Update any claims that no longer match reality.

Authoritative source order

When files contradict each other: actual code > governance docs > CLAUDE.md/AGENTS.md > tool-specific memories > planning docs.

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 · 33 lines · 0 tokens per session scan A 11ef56925097

Subscribe to this mod's changes

ctx-audit is a cursor rule published in the GitHub repository knitli/toolshed (1 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 336 tokens. 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.