security-reviewer

security-reviewer is an agent for coding agents from cheezy/stride-security-review. It costs 489 tokens per session (17,475 once invoked), scanned D, original, MIT.

An AI application-security reviewer that examines changed code, whole files, or changes against a list of security requirements.

In plain words
What is it for?
Use it to review code diffs, inspect complete files, or verify whether each security requirement in a task was addressed.
Why use it?
It looks at how data moves through the surrounding code, helping identify realistic vulnerabilities that simple text searches can miss.

Agent

Part of the stride-security-review plugin — 1 skill, 1 command, 1 agent shipped together

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/cheezy/stride-security-review/security-reviewer
Clone the repo
git clone --depth 1 https://github.com/cheezy/stride-security-review

Or install stride-security-review, the plugin that ships this one along with the rest of its 1 skill, 1 command, 1 agent.

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 security-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/cheezy/stride-security-review/security-reviewer.svg)](https://agentmods.dev/agents/cheezy/stride-security-review/security-reviewer)
Your own site
<a href="https://agentmods.dev/agents/cheezy/stride-security-review/security-reviewer"><img src="https://agentmods.dev/badge/agents/cheezy/stride-security-review/security-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 489 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 17,475 The whole file, excluding the scripts and references it only reads on demand.
Security scan D 5 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 $0.00489 $0.17475
Opus 5 $0.00244 $0.08738
Sonnet 5 $0.00098 $0.03495
Haiku 4.5 $0.00049 $0.01748

Measured 4d ago against content hash 2a60e10d5aaa, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade D, and why

security-reviewer scanned grade D with 5 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.

Downloads and executes remote codemediumSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

| **Pipe-to-shell installer** | Any shell construct that fetches a remote payload and executes it without verification. Patterns: `curl ... \| sh`, `curl ... \| bash`, `wget ... \| sh`, `wget -O- ... \| bash`, PowerShell

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

Cloud metadata endpointmediumServer-side request forgery

One request to 169.254.169.254 can return temporary IAM credentials.

| SSRF via `requests` / `urllib` / `httpx` with user URL | input_validation | A `requests.get(url)` / `requests.post(url, ...)`, `urllib.request.urlopen(url)`, or `httpx.AsyncClient().get(url)` call where `url` traces to

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.

| `child_process.exec` / `execSync` with user input | injection | `child_process.exec(\`convert ${req.body.path} out.png\`)`, `execSync('rm -rf ' + req.query.dir)`, or `spawn('sh', ['-c', interp])` where the command stri

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.

| SSRF via `requests` / `urllib` / `httpx` with user URL | input_validation | A `requests.get(url)` / `requests.post(url, ...)`, `urllib.request.urlopen(url)`, or `httpx.AsyncClient().get(url)` call where `url` traces to

Runs shell commandslowCapability

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

| `child_process.exec` / `execSync` with user input | injection | `child_process.exec(\`convert ${req.body.path} out.png\`)`, `execSync('rm -rf ' + req.query.dir)`, or `spawn('sh', ['-c', interp])` where the command stri
agents/security-reviewer.md · 388 lines

How it starts

The opening of the file, as written. The whole thing — 388 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are a senior application security reviewer. Your job is to analyze code for security vulnerabilities and return a structured list of findings. You favor semantic analysis over pattern matching — what makes you better than grep is that you understand the control flow, the data flow, and the surrounding context before flagging an issue.

You receive input in one of three modes (see Input modes below): a unified diff against HEAD; one or more whole files when running in full_file mode; or a diff paired with the task's security_considerations list when running in considerations mode. The caller tells you which. The analysis methodology, vulnerability classes, severity rubric, and realism filter are the same in every mode — only the unit of review changes, and considerations mode adds a single output field (consideration_verdicts).

Analysis methodology

For every region you review — a changed hunk in diff mode, or each file in full_file mode — ask in order:

  1. What is the trust boundary? Is this code receiving data from an untrusted source (user input, network, file system, environment) and producing an effect (database query, shell command, response, cryptographic operation, redirect)? If not — and the data stays internal — most vulnerability classes do not apply.
  2. Is the data path correctly defended? Look for parameterization, escaping, allow-listing, canonicalization, authentication checks, authorization checks, and constant-time comparisons. The absence of a defense at a trust boundary is a finding; the presence of one is not.
  3. Is the defense actually doing what its name implies? Calling a function named escape does not mean output is safe — verify the function escapes for the correct sink. A SQL-escape applied to an HTML sink is still XSS.
  4. What's the worst realistic outcome if a finding is genuine? Use this to assign severity. Do not flag a finding whose realistic worst case is "minor information disclosure of metadata an attacker could obtain elsewhere".

Read the full file on GitHub · 388 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. 4d ago First seen · 388 lines · 489 tokens per session scan D 2a60e10d5aaa

Subscribe to this mod's changes

security-reviewer is an agent published in the GitHub repository cheezy/stride-security-review (5 stars, last pushed 13d ago), licensed MIT. It adds 489 tokens to every session and 17,475 once invoked, about $0.0024 per session on Opus 5. A static security scan graded it D with 5 findings (downloads and executes remote code, cloud metadata endpoint, recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.