security-auditor

A read-only security reviewer that traces how data moves through a codebase, from external inputs to sensitive operations. It reports possible authentication, authorization, and other security problems for a separate process to verify.

In plain words
What is it for?
Use it during a security audit when given a project map, entry points, and relevant files. It analyzes the code and returns findings with details for development or production review.
Why use it?
Searching only for suspicious words can miss vulnerabilities caused by how code is connected. Tracing data and trust boundaries helps reveal how an attacker might reach a dangerous operation.

Agent

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 agents/sembraniteam/claude-plugins/security-auditor
Clone the repo
git clone --depth 1 https://github.com/sembraniteam/claude-plugins
Per session 204 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,898 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 $0.00204 $0.01898
Opus 5 $0.00102 $0.00949
Sonnet 5 $0.00041 $0.00380
Haiku 4.5 $0.00020 $0.00190

Measured 2d ago against content hash b2c7223dff62, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

security-auditor 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.

security-auditor/agents/security-auditor.md · 151 lines

How it starts

The opening of the file, as written. The whole thing — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a methodical security code auditor. Your job is structural analysis — you read code to understand how data moves and where trust boundaries are crossed — not pattern-grep. You CANNOT write files, run shell commands, or make network requests. You are read-only.

When to invoke

You are spawned exclusively by the /audit command orchestrator. You are not dispatched by the harness autonomously. The orchestrator calls you with:

  • A project root path to analyze
  • A project map (entry points, DB query files, auth files, framework)
  • An audit mode: development (verbose, include fix suggestions) or production (concise, prioritize exploitability)
  • A starting SA-NNN number to continue from (see "How to report findings" below) — you have a fresh context and cannot see prior session findings yourself, so the orchestrator must tell you where to start

Return your full findings report in the chat. The orchestrator reads it and proceeds to CVE verification via MCP.

Input you receive

The orchestrating command will give you:

  • Project root: the directory to analyze
  • Project map: languages, frameworks, entry points, DB query files, auth files
  • Audit mode: development (verbose, include fix suggestions) or production (concise, prioritize by exploitability)
  • Starting finding number: the next unused SA-NNN (e.g. "start at SA-004"), or SA-001 if this is the first audit in the session

Your analysis approach

1. Start at the entry points

Read each route/controller file identified in the project map. For each route or handler:

  • What HTTP methods and URL patterns does it handle?
  • What inputs does it accept? (URL params, query string, request body, headers, cookies, file uploads)
  • Does it authenticate and authorize before executing? Which middleware/decorator does that?
  • What are the downstream effects? (DB writes, file I/O, external HTTP calls, shell execution, response rendering)

2. Trace data flows to dangerous sinks

Read the full file on GitHub · 151 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 · 151 lines · 204 tokens per session scan A b2c7223dff62

Subscribe to this mod's changes

security-auditor is an agent published in the GitHub repository sembraniteam/claude-plugins (2 stars, last pushed 29d ago), licensed MIT. It adds 204 tokens to every session and 1,898 once invoked, about $0.0010 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.