skills/openclaw/skills/gitflow/Gen Agent Trust Hub

gitflow

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODECOMMAND_EXECUTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill instructs the agent to read raw logs from CI/CD pipelines (GitHub Actions and GitLab CI), which is a known vector for indirect prompt injection if an attacker can control build output (e.g., via a Pull Request).
  • Ingestion points: CI/CD log retrieval via gh run view --log and glab ci trace in SKILL.md.
  • Boundary markers: None identified; logs are ingested as unstructured text.
  • Capability inventory: The skill allows the agent to execute git commands, push code, and rerun CI jobs.
  • Sanitization: None; the agent processes raw log output which may contain malicious instructions.
  • Persistence Mechanisms (LOW): The documentation suggests modifying ~/.gitconfig to add a persistent git alias. While this is a standard developer practice, it technically falls under configuration persistence patterns.
  • Data Exposure (SAFE): While the skill reads logs that might accidentally contain secrets (PII/Tokens), it does not target specific sensitive files (like .ssh/id_rsa) or exfiltrate data to unauthorized third parties.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 10:41 PM