Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
SSkills DirectorySkills Directory
Is that skill safe to install?
Are you the author of Security Review?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/reliant-labs-security-review)
---
name: security-review
description: Security review methodology — vulnerability detection, injection analysis, authentication and access-control verification, and structured security reporting
---
# Security Review
## Focus Areas
### Injection Vulnerabilities
- **SQL injection**: parameterized queries vs string concatenation, ORM misuse
- **Shell/command injection**: user input in exec/system calls, argument injection
- **Template injection**: server-side template engines with user input
- **Regex injection**: ReDoS via user-controlled patterns
- **Path traversal**: `../` sequences, symlink attacks, archive extraction (zip slip)
- **Command argument injection**: flags/options parsed from user input
### Authentication vs Authorization (AuthN vs AuthZ)
**Forge provides authentication only — it does not provide authorization.** No
framework layer decides whether a caller may act; the application implements
and enforces that itself, in whatever shape the product needs. So authenticating
a request proves nothing about whether it was allowed. Review accordingly: for
every protected endpoint, find the check that authorizes the *action* and the
*specific records* touched. An endpoint that only establishes identity and then
acts is a finding, not a pass.
- **Permission checks**: missing or incorrect authorization on endpoints/functions
- **Data scoping**: cross-account data access, missing authorization filters in queries
- **Object-level access control**: can user A access user B's resources?
- **Privilege escalation**: can regular users access admin functionality?
- **Session management**: fixation, hijacking, improper invalidation
### Secrets & Sensitive Data
- **Logged accidentally**: secrets in error messages, debug logs, stack traces
- **Stored in configs**: hardcoded in source, committed to git
- **Sent to client**: secrets in API responses, frontend bundles, error details
- **Leaked in errors**: stack traces exposing internal paths, versions, configs
- **Env var handling**: secrets readable by child processes, in crash dumps
### Input Validation & Trust Boundaries
- **Trust boundary violations**: treating external input as trusted
- **Canonicalization issues**: URL normalization, path resolution, unicode handling
- **Size limits**: missing limits enabling DoS (large payloads, many items)
- **Type confusion**: unexpected types bypassing validation
- **Deserialization**: unsafe deserialization of user-controlled data
### Cryptography
- **Homegrown crypto**: custom encryption, hashing, or signing implementations
- **Weak randomness**: math/rand instead of crypto/rand, predictable seeds
- **Missing constant-time comparison**: timing attacks on secret comparison
- **Deprecated algorithms**: MD5, SHA1 for security purposes, weak ciphers
- **Key management**: hardcoded keys, insufficient key rotation, weak key derivation
## Output Format
### Security Review Summary
- Overall risk assessment: LOW / MEDIUM / HIGH / CRITICAL
- Number of findings by severity
### Findings (ordered by severity)
For each finding:
- **Severity**: CRITICAL / HIGH / MEDIUM / LOW / INFO
- **Category**: (e.g., "SQL Injection", "Missing AuthZ")
- **Location**: file:line or function name
- **Description**: What the vulnerability is
- **Attack scenario**: How it could be exploited
- **Recommendation**: Specific fix
### Secure Patterns Observed
- Note any good security practices in the code (brief)
## Review Guidelines
- Assume all external input is malicious
- Check the full data flow, not just the immediate code
- Consider what happens when dependencies fail or return unexpected data
- Look for missing checks, not just incorrect checks
- Flag security-relevant TODOs or FIXMEs
- Keep your review proportional to the change. A CSS change or copy update does not warrant a security deep-dive.