Security Scanner
AI-powered pentest simulation and compliance check for your project.
What it does
The Security Scanner analyzes your project's actual codebase for vulnerabilities and generates a structured security report with severity ratings, CWE references, and actionable fixes. It runs entirely locally on your machine — no code leaves your system.
The Security tab is available once you've set a source path for the idea. See the Coder Agent guide for instructions on linking your project folder.
Pentest Simulation
A source path must be set before the Security tab is active.
The AI scans your project files for common vulnerability classes.
Each finding includes severity level, CWE reference, affected file and line number, and a concrete recommendation.
Vulnerability classes scanned:
- OWASP Top 10 (injection, broken auth, sensitive data exposure, etc.)
- Injection vulnerabilities (SQL, command, LDAP)
- Authentication and session management issues
- Exposed secrets and hardcoded credentials
- Insecure direct object references
- Security misconfiguration
Compliance Check
Run a gap analysis against common security standards. The compliance check reviews your codebase and configuration against:
- OWASP ASVS — Application Security Verification Standard, Level 1 and Level 2
- GDPR basics — data handling, consent, storage, and deletion requirements
- API security best practices — rate limiting, authentication, input validation, CORS
Results are grouped by standard and show which controls pass, which fail, and what action is needed to close each gap.
Auto-Fix
For supported finding types, an Auto-fix button appears alongside the finding. Clicking it hands the finding to the Coder Agent, which generates a patch.
- Review the diff before approving — same approve/reject flow as the Coder Agent
- Auto-fix is available for common patterns: SQL injection, hardcoded secrets, missing input validation, etc.
- Complex architectural issues may not have an auto-fix — you'll see guidance instead
Finding severities
| Severity | Meaning | Action |
|---|---|---|
| Critical | Directly exploitable in its current state | Fix immediately — before any public exposure |
| High | Significant risk, likely exploitable with moderate effort | Fix before launch |
| Medium | Real risk but requires specific conditions to exploit | Fix in your next sprint |
| Low | Best-practice improvement, low exploit risk | Address when convenient |