What is security-audit?

A staged security audit of the code in the current working directory, looking for real vulnerabilities that an attacker could exploit.

Use it to map the attack surface, investigate security risks in parallel, write proof-of-concept tests, apply verified fixes, and produce a final audit report.

It separates planning, investigation, verification, and reporting so security findings can be checked and ranked instead of being mixed with style feedback.

security-audit — Command for Claude Code by synadia-ai: A staged security audit of the code in the…

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/synadia-ai/synadia-agents/security-audit

Clone the repo

git clone --depth 1 https://github.com/synadia-ai/synadia-agents

Made for: Claude Code.

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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/synadia-ai/synadia-agents/security-audit.svg)](https://agentmods.dev/commands/synadia-ai/synadia-agents/security-audit)

Your own site

<a href="https://agentmods.dev/commands/synadia-ai/synadia-agents/security-audit"><img src="https://agentmods.dev/badge/commands/synadia-ai/synadia-agents/security-audit.svg" alt="Measured on agentmods" height="20"></a>

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 7,075 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 4 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.

Model	Per session	Once invoked
Fable 5	$0.00000	$0.07075
Opus 5	$0.00000	$0.03537
Sonnet 5	$0.00000	$0.01415
Haiku 4.5	$0.00000	$0.00707

Measured 4d ago against content hash 0ffc1bf91317, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

security-audit scanned grade B with 4 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 4d 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.

Unrestricted tool accesslowExcessive agency

A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.

- Could a compromised dependency execute arbitrary code at build time?

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

Recursive force deletemediumDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

After the report is written, clean up: `rm -rf .security-audit/` (the intermediate files are not meant to be committed).

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

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

- A curl/CLI command that triggers the issue

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

- Command injection: shell execution with user input? (search for exec, system, popen, subprocess, child_process)

.claude/commands/security-audit.md · 588 lines

How it starts

The opening of the file, as written. The whole thing — 588 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Full Security Audit

You are performing a comprehensive security audit of the codebase in the current working directory. Treat this with the rigor of a professional penetration test. Your job is to find real, exploitable vulnerabilities - not style issues.

Execution Strategy

You MUST use subagents extensively to avoid context exhaustion. Findings are passed between phases via files (in .security-audit/), NOT via the main agent's context. The main agent only sees TL;DR summaries.

The audit has 5 phases:

Planning - Discover project structure, identify attack surface
Deep Investigation - Parallel subagents per security domain → write full findings to .security-audit/findings/
Initial Report & Triage - Synthesize TL;DR summaries, rank by severity → write triage index to .security-audit/triage.md
Verification & Fix - Parallel subagents read finding files, verify, write PoC tests and fixes → write to .security-audit/findings/SEC-*-verified.md
Final Report - Assemble from all files → write SECURITY_AUDIT_REPORT.md

After the report is written, clean up: rm -rf .security-audit/ (the intermediate files are not meant to be committed).

PHASE 1: PLANNING & DISCOVERY

Before spawning any investigation agents, you must understand what you're auditing. Do this yourself (do NOT delegate):

Step 1: Discover the project

Read CLAUDE.md, AGENTS.md, README.md, CONTRIBUTING.md if they exist
Read the primary manifest file(s): Cargo.toml, package.json, go.mod, pyproject.toml, pom.xml, build.gradle, CMakeLists.txt, Makefile, etc.
Run ls on the root and src/ (or equivalent) to understand the directory layout
Identify the language(s), framework(s), build system, and project type (library, server, CLI tool, web app, etc.)

Step 2: Identify security-relevant areas

Based on what you discover, identify:

Trust boundaries: Where does external input enter? (network, files, CLI args, environment variables, IPC, database, user input in web forms, API requests)
Attack surface: Which modules handle untrusted data? (parsers, protocol handlers, auth, crypto, serialization/deserialization, file I/O, SQL/query builders, template rendering, HTTP handlers)
Threat actors: Who could attack this? (remote unauthenticated, authenticated user, malicious peer/server, man-in-the-middle, local attacker, supply chain)
Security-critical files: List the specific files/modules that handle auth, crypto, parsing, network I/O, input validation, and access control

Read the full file on GitHub · 588 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.

4d ago First seen · 588 lines · 0 tokens per session scan B 0ffc1bf91317

Subscribe to this mod's changes

security-audit is a command published in the GitHub repository synadia-ai/synadia-agents (76 stars, last pushed 5d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 7,075 tokens. A static security scan graded it B with 4 findings (unrestricted tool access, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

LLM Security agent-protocol ai ai-agents claude-code nats

security-audit