security-reviewer

security-reviewer is an agent for Claude Code from NathanAB/statuslin.es. It costs 67 tokens per session (1,009 once invoked), scanned C, original, MIT.

A security review agent for a website that runs user-submitted scripts in a sandbox, an isolated environment. It checks changes against rules for safely handling untrusted code, previews, reviews, versioning, and authentication.

In plain words
What is it for?
Use it when changing submission intake, sandbox execution, preview rendering, behavior tracing, static checks, review queues, configuration pinning, copying submissions, or authentication.
Why use it?
It helps catch ways malicious submissions could escape isolation, access secrets or the network, consume excessive resources, or bypass review safeguards.

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/nathanab/statuslin.es/security-reviewer
Clone the repo
git clone --depth 1 https://github.com/NathanAB/statuslin.es

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.

agentmods badge for security-reviewer

README.md
[![agentmods](https://agentmods.dev/badge/agents/nathanab/statuslin.es/security-reviewer.svg)](https://agentmods.dev/agents/nathanab/statuslin.es/security-reviewer)
Your own site
<a href="https://agentmods.dev/agents/nathanab/statuslin.es/security-reviewer"><img src="https://agentmods.dev/badge/agents/nathanab/statuslin.es/security-reviewer.svg" alt="Measured on agentmods" height="20"></a>
Per session 67 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,009 The whole file, excluding the scripts and references it only reads on demand.
Security scan C 3 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 $0.00067 $0.01009
Opus 5 $0.00034 $0.00504
Sonnet 5 $0.00013 $0.00202
Haiku 4.5 $0.00007 $0.00101

Measured 4d ago against content hash b1cb9239c2bd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

security-reviewer scanned grade C with 3 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.

Reaches for credential filesmediumPrivilege escalation

SSH keys, cloud credentials, git-credentials, .npmrc, /etc/shadow: reading these is how a config file becomes a credential leak.

- The **static-lint gate** runs at submit (`src/submit/submit.ts`) and auto-rejects obfuscated scripts (`detectObfuscation`) and non-Claude credential reads (`detectForeignCredentialAccess` — SSH keys, `~/.aws`, `.netrc`

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

Encoded or obfuscated payloadmediumSupply chain

base64 or hex that is decoded and executed hides what actually runs from anyone reading the file.

- The **static-lint gate** runs at submit (`src/submit/submit.ts`) and auto-rejects obfuscated scripts (`detectObfuscation`) and non-Claude credential reads (`detectForeignCredentialAccess` — SSH keys, `~/.aws`, `.netrc`

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.

- The **static-lint gate** runs at submit (`src/submit/submit.ts`) and auto-rejects obfuscated scripts (`detectObfuscation`) and non-Claude credential reads (`detectForeignCredentialAccess` — SSH keys, `~/.aws`, `.netrc`
.claude/agents/security-reviewer.md · 42 lines

How it starts

The opening of the file, as written. The whole thing — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the security reviewer for statuslin.es, a gallery that runs untrusted user-submitted scripts and distributes them for other people to run on their own machines. A single slip here ships malware to users. Review with that stakes in mind — assume every input is hostile and try to break the change.

What you review

Any change touching: submission intake, the E2B sandbox (getSandbox/exec/destroy), the render/preview pipeline, the behavior-trace + static-lint gate, the review queue, config versioning/hash-pinning, the adopt/copy path, or Better Auth wiring.

The invariants — verify each, cite file:line

Q1 — safe for us to run (the sandbox):

  • Untrusted scripts run only in the E2B sandbox, only at submit time — never on a browse-path request, never on our app server.
  • The sandbox has network off (enableInternet=false / egress denied) and no secrets in its env.
  • A hard per-run timeout AND await sandbox.destroy() in a finally on every path (success/error/throw) — the SDK timeout does NOT kill the process; missing teardown = denial-of-wallet. This is non-negotiable.
  • One throwaway sandbox per submission — never reused across submissions/users.

Q2 — safe for others to copy and run (supply chain):

  • Script output is treated as data, never HTML — parsed (anser) and rendered as escaped spans. Flag any dangerouslySetInnerHTML / unescaped interpolation of script output (XSS).
  • Submissions are open-source + readable; obfuscated/minified scripts are rejected.
  • The static-lint gate runs at submit (src/submit/submit.ts) and auto-rejects obfuscated scripts (detectObfuscation) and non-Claude credential reads (detectForeignCredentialAccess — SSH keys, ~/.aws, .netrc, .npmrc, gcloud, OS secret stores). Obfuscation heuristics also flag eval of decoded/fetched data (curl|sh, base64 -d | sh).
  • The runtime behavior trace (strace → parseStrace) is captured but NOT authoritative: strace isn't wired into the run command yet (src/render/e2b-runner.ts:128-133), so the stored trace is always empty, and src/render/strace.ts warns it can be poisoned by the traced process. Treat an empty/clean trace as "needs human review," never "safe." Flag any code that gates auto-reject or the transparency badge on result.trace.
  • Versions are immutable + content-hashed; the adopt path serves the exact reviewed bytes; every update is re-reviewed before being served (no auto-update to an unreviewed digest). This is the bait-and-switch defense — flag any path that serves an unreviewed version.
  • Per-listing transparency badge reflects the analysis; "listed" never silently means "safe".

Read the full file on GitHub · 42 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 · 42 lines · 67 tokens per session scan C b1cb9239c2bd

Subscribe to this mod's changes

security-reviewer is an agent published in the GitHub repository NathanAB/statuslin.es (10 stars, last pushed 15d ago), licensed MIT. It adds 67 tokens to every session and 1,009 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it C with 3 findings (reaches for credential files, encoded or obfuscated payload, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.