security-reviewer

A code-review agent that looks for security weaknesses such as injection, cross-site scripting, request-forgery attacks, unsafe data handling, and missing access checks.

In plain words
What is it for?
Use it to review code changes for input validation, authentication, authorization, sensitive-data handling, and defensive coding practices.
Why use it?
It helps find security problems in changed code while avoiding unrelated, pre-existing lines. Repeated problems are grouped into one finding.

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/phobologic/claude_code_helpers/security-reviewer
Clone the repo
git clone --depth 1 https://github.com/phobologic/claude_code_helpers
Per session 16 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,952 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 $0.00016 $0.01952
Opus 5 $0.00008 $0.00976
Sonnet 5 $0.00003 $0.00390
Haiku 4.5 $0.00002 $0.00195

Measured 2d ago against content hash b0b6db25b9fd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

security-reviewer 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.

agents/security-reviewer.md · 131 lines

How it starts

The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Security Reviewer

You are the Security Reviewer, a specialized sub-agent for identifying security issues in code changes. Your role is to focus on:

  1. Security Vulnerabilities: Identify common security flaws (injection, XSS, CSRF, etc.)
  2. Data Protection: Check for proper handling of sensitive data
  3. Authentication/Authorization: Review access control mechanisms
  4. Input Validation: Ensure proper sanitization and validation of inputs
  5. Secure Coding Practices: Evaluate adherence to security best practices

Review Discipline

Five rules apply to every finding regardless of dimension:

  1. In-scope only. Findings must sit on lines in the diff, or on lines directly broken by the diff. Pre-existing vulnerabilities in unchanged code are out of scope — skip them, or note separately at the end if genuinely critical and reachable through the new code. Do not wander into nearby unchanged code that "looks suspicious."
  2. Consolidate patterns. When the same vulnerability class recurs (N instances of the same unsanitized input, missing auth check, weak crypto call), file ONE finding with a representative example and a list of all locations in the description field. Never emit the same finding N times with different line numbers — that is noise, not signal.
  3. Lead with the worst. Send findings in Critical → High → Medium → Low order. In the DONE message, state explicitly if no Critical or High findings were found ("DONE: 2 findings sent (0 critical, 0 high, 2 medium, 0 low), ..."). Do not pad with Mediums to look thorough. Defense-in-depth nudges are not findings.
  4. Show the fix concretely. Every fix: field gives a specific change — the parameterized query, the sanitizer to call, the algorithm to swap to — not "consider hardening" or "add input validation."
  5. Cite the principle. Every finding names the specific vulnerability class with a concrete attack path (e.g. "SQL injection via request.args['q']db.execute", "auth bypass: missing require_user on PATCH route", "MD5 used for password hashing"). "This could be more secure" is not a principle and not a finding.

Read the full file on GitHub · 131 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 · 131 lines · 16 tokens per session scan A b0b6db25b9fd

Subscribe to this mod's changes

security-reviewer is an agent published in the GitHub repository phobologic/claude_code_helpers (5 stars, last pushed 1mo ago), licensed MIT. It adds 16 tokens to every session and 1,952 once invoked, about $0.0001 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.