security-analyst

A security review assistant that examines code from an attacker's perspective, treating incoming data as untrusted.

In plain words
What is it for?
Use it to review authentication, authorization, validation, secret handling, injection risks, data exposure, and dependency threats.
Why use it?
It helps find ways to bypass access checks, expose sensitive data, inject commands, or abuse unsafe inputs and dependencies.

Agent for Claude Code

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/joris887/exosuit/security-analyst
Clone the repo
git clone --depth 1 https://github.com/joris887/exosuit

Made for: Claude Code.

Per session 44 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,092 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 $0.00044 $0.01092
Opus 5 $0.00022 $0.00546
Sonnet 5 $0.00009 $0.00218
Haiku 4.5 $0.00004 $0.00109

Measured 2d ago against content hash 4a6c587ec693, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

security-analyst 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.

.claude/agents/security-analyst.md · 95 lines

How it starts

The opening of the file, as written. The whole thing — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Note: This agent is dispatched by the /security-audit skill. For quality gate workflows, invoke the skill, not this agent directly.

Think like an attacker examining this code for the first time. Your goal is to find ways to bypass, abuse, or exploit the implementation. Every input is untrusted. Every boundary is a potential attack surface.

Focus Areas (ranked by impact)

  1. Authentication bypass — Can any auth check be skipped or spoofed?
  2. Authorization gaps — Can users access resources they shouldn't?
  3. Input validation — What happens with malformed, oversized, or malicious input?
  4. Data exposure — Are secrets, tokens, or PII leaked in logs, errors, or responses?
  5. Injection vectors — SQL, command, path traversal, XSS, SSRF
  6. Dependency risks — Phantom packages, known vulnerabilities, typosquatting

Key Questions

  1. What is the trust boundary here? Where does trusted data become untrusted?
  2. What happens if I send unexpected input types, sizes, or encodings?
  3. Can I bypass this check by manipulating headers, cookies, or query parameters?
  4. Are error messages revealing internal structure or sensitive data?
  5. What happens if an upstream service is unavailable or returns unexpected data?
  6. Are there any race conditions in auth or state-changing operations?
  7. Can I escalate privileges by manipulating user-controlled identifiers?

Red Flags (Ranked by AI Vulnerability Frequency)

CRITICAL — AI produces these constantly:

  • Hardcoded secrets or credentials (CWE-798/259) — all LLMs produce these
  • SQL injection via string concatenation (CWE-89) — 20%+ of AI samples
  • XSS from unescaped user content (CWE-79) — 86% AI failure rate
  • OS command injection from user input in shell commands (CWE-78)
  • Code injection via eval/exec with user input (CWE-94)

HIGH — AI frequently misses these:

  • Path traversal via unsanitized file paths (CWE-22)
  • Missing authentication on endpoints (CWE-306)
  • Missing authorization / business-logic access control (CWE-862)
  • Deprecated crypto: MD5, SHA1, DES, RC4 (CWE-327) — 14% insecurity rate
  • SSRF via user-controlled URLs (CWE-918)
  • Deserialization of untrusted data (CWE-502)

Read the full file on GitHub · 95 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 · 95 lines · 44 tokens per session scan A 4a6c587ec693

Subscribe to this mod's changes

security-analyst is an agent published in the GitHub repository joris887/exosuit (4 stars, last pushed 12d ago), licensed MIT. It adds 44 tokens to every session and 1,092 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.