conflict-resolution

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • PROMPT_INJECTION (HIGH): The skill is highly susceptible to indirect prompt injection due to its core functionality.
  • Ingestion points: The skill reads external content from files with conflict markers (e.g., cat src/services/user.ts, git status).
  • Boundary markers: It relies on standard Git conflict markers (<<<<<<<, =======, >>>>>>>) which are not robust security boundaries against adversarial content.
  • Capability inventory: The skill can write to the filesystem (git checkout, git add), commit code, and execute arbitrary code via the test runner (pnpm test).
  • Sanitization: There is no logic to sanitize or filter the content of the code being merged, allowing embedded instructions to potentially influence the agent's behavior during the resolution process.
  • COMMAND_EXECUTION (MEDIUM): The skill executes multiple system commands with side effects.
  • Evidence: It uses git push --force-with-lease, which can rewrite remote history, and pnpm test, which executes project-defined scripts that could contain malicious code if the project files themselves were compromised via a conflict injection.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 05:58 AM