mpm-postmortem

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious instructions, obfuscation, or data exfiltration patterns were detected in the skill configuration.
  • [NO_CODE]: The file serves as metadata and usage documentation for an external command; it does not contain executable logic or script content.
  • [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection by design, as it ingests untrusted data (error logs from scripts and user code) to generate automated fixes.
  • Ingestion points: Error logs originating from scripts, skills, agents, and user code (SKILL.md).
  • Boundary markers: No delimiters or instructions to ignore embedded content are specified in this documentation.
  • Capability inventory: The skill supports high-privilege operations including --auto-fix (file modification) and --create-prs (remote repository updates).
  • Sanitization: No sanitization or validation steps for the analyzed error content are described in the provided file.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 02:30 PM
Security Audit — agent-trust-hub — mpm-postmortem