codex-claude-loop

Warn

Audited by Snyk on Aug 24, 2026

Risk Level: MEDIUM
Full Analysis

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

  • Third-party content exposure detected (medium risk: 0.30). The workflow’s runtime sends a user-authored task plan and implementation text into codex exec via echo "...[Claude's plan here]..." | codex exec ... and echo "Review this implementation..." | codex exec ..., so outsider free text flows into the validator without selecting a specific item.

Issues (1)

W011
MEDIUM

Third-party content exposure detected (indirect prompt injection risk).

Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 24, 2026, 12:52 PM
Issues
1
Security Audit — snyk — codex-claude-loop