github-automation

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it processes untrusted data and has significant system influence.
  • Ingestion points: The skill ingests untrusted content from external sources via GITHUB_LIST_REPOSITORY_ISSUES, GITHUB_SEARCH_CODE, and GITHUB_GET_REPOSITORY_CONTENT.
  • Boundary markers: There are no instructions or prompt templates provided to delimit external data or instruct the agent to ignore instructions embedded within issues, PRs, or code files.
  • Capability inventory: The skill possesses extensive administrative capabilities, including GITHUB_MERGE_A_PULL_REQUEST, GITHUB_DELETE_A_REPOSITORY, GITHUB_UPDATE_BRANCH_PROTECTION, and GITHUB_ADD_A_REPOSITORY_COLLABORATOR.
  • Sanitization: The skill lacks specified mechanisms for validating or sanitizing data retrieved from GitHub repositories before it is processed by the agent.
  • [COMMAND_EXECUTION]: The skill provides tools for high-impact repository operations that could lead to data loss or unauthorized access if misused.
  • Evidence: The skill documentation includes tools for permanent repository deletion (GITHUB_DELETE_A_REPOSITORY), modifying security settings (GITHUB_UPDATE_BRANCH_PROTECTION), and changing access control (GITHUB_ADD_A_REPOSITORY_COLLABORATOR). Although the skill recommends user confirmation, the presence of these tools in a single context increases the risk of autonomous or coerced misuse.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 08:24 AM
Security Audit — agent-trust-hub — github-automation