cvp-defense-audit

cvp-defense-audit is a skill for Claude Code from charlesjones-dev/claude-code-plugins-dev. It costs 115 tokens per session (6,245 once invoked), scanned A, original, MIT.

A planning guide for a defensive security audit of an authorized code repository. It prepares a sandboxed audit session, where the model searches for vulnerabilities, proves findings with failing tests, and later helps verify and fix them.

In plain words
What is it for?
Use it to map a repository's attack surface, create an audit prompt and launch checklist, and prepare a fix phase with regression tests. A regression test is a test that prevents a repaired problem from returning.
Why use it?
It separates investigation from code changes, making it easier to review evidence before repairs are made. It also keeps audit test files in a designated ignored folder.

Skill for Claude Code

Written for Claude Code: allowed-tools in frontmatter. Also seen: reads .claude/ paths; mentions CLAUDE.md; mentions subagents.

Part of the ai-cvp plugin — 1 skill shipped together

Good fit Use it to map a repository's attack surface, create an audit prompt and launch checklist, and prepare a fix phase with regression tests. A regression test is a test that prevents a repaired problem from returning.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit
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.

Any agent
npx skills add charlesjones-dev/claude-code-plugins-dev --skill cvp-defense-audit
Clone the repo
git clone --depth 1 https://github.com/charlesjones-dev/claude-code-plugins-dev

Made for: Claude Code.

Or install ai-cvp, the plugin that ships this one along with the rest of its 1 skill.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for cvp-defense-audit

README.md
[![agentmods](https://agentmods.dev/badge/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit/github.svg)](https://agentmods.dev/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit)
Your own site
<a href="https://agentmods.dev/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit"><img src="https://agentmods.dev/badge/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for cvp-defense-audit

Your own site · 80×15
<a href="https://agentmods.dev/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit"><img src="https://agentmods.dev/badge/skills/charlesjones-dev/claude-code-plugins-dev/cvp-defense-audit.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 115 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 6,245 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00115 $0.06245
Opus 5.5 $0.00046 $0.02498
Sonnet 5.5 $0.00023 $0.01249
Haiku 4.5 $0.00012 $0.00624

Measured 7d ago against content hash c8d81a1c1286, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

cvp-defense-audit scanned grade A with 1 finding 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 7d 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.

Reads agent configuration directorieslowAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

3. Ask Claude to run `touch ~/sandbox-probe`. Expect `Operation not permitted` (macOS) or `Read-only file system` (Linux, WSL2). If it succeeds, delete the file and quit: a settings file is widening the sandbox, so check

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

plugins/ai-cvp/skills/cvp-defense-audit/SKILL.md · 216 lines

How it starts

The opening of the file, as written. The whole thing — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CVP defense audit

Use this only on code the user owns or is authorized to test.

This skill is phase 1 of a three-phase defensive audit:

  1. Plan (this skill). A normal, unsandboxed session on an Opus-class planning model does read-only recon, makes sure git ignores docs/security/, and writes a paste-ready audit prompt, a sandbox launch checklist, and a fix prompt.
  2. Audit. The user starts a new session on the strongest security-capable model they can use (the audit model), launched in auto mode with a strict sandbox set on the command line, and pastes the audit prompt. The audit model maps the attack surface, hunts, proves each finding with a failing test, and writes a report. The prompt ends with an output contract: the report at a fixed path, each test stored as a disabled file under docs/security/tests/, anything else it keeps under docs/security/harness/, and a check to run before finishing. Everything it leaves behind is under docs/security/, which git ignores. It fixes nothing.
  3. Fix. Back in a normal, unsandboxed session on an Opus-class fixing model, the user pastes the fix prompt. The fixing model verifies each proof, settles what code alone couldn't answer, lets the user pick what to fix, and ships each fix with its archived proof test restored as the regression test.

Do the recon here so the audit model spends its budget hunting and proving bugs instead of orienting. The supporting files sit next to this one, in ${CLAUDE_SKILL_DIR}.

Context

  • Date: !date +%F
  • Repo root: !git rev-parse --show-toplevel 2>/dev/null || pwd
  • Branch: !git branch --show-current 2>/dev/null || echo none
  • Remote: !git remote get-url origin 2>/dev/null || echo none
  • Last commit: !git log -1 --format='%h %s (%cr)' 2>/dev/null || echo none

Without git, the repo root is the working directory and the other values read none. Without a remote, skip the GitHub reads in step 2.

Arguments (may be empty): $ARGUMENTS

  • --model <id> (or --model=<id>) is the audit model ID for the launch line.
  • The rest is an optional focus: an area or path, such as billing or src/auth.
  • If the arguments also state the user's authorization, such as "enrolled in the Cyber Verification Program with Defense Access" or "authorized pentest for , scope: api/ only", use it in step 1 instead of asking.

Read the full file on GitHub · 216 lines

Files

What ships with it

3 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. 7d ago Changed · +13 lines c8d81a1c1286
  2. 9d ago First seen · 203 lines · 115 tokens per session scan A 3d5f9a218c0c

Subscribe to this mod's changes

cvp-defense-audit is a skill published in the GitHub repository charlesjones-dev/claude-code-plugins-dev (35 stars, last pushed 6d ago), licensed MIT. It adds 115 tokens to every session and 6,245 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-08.