skills/thaitype/chief/chief-upgrade/Gen Agent Trust Hub

chief-upgrade

Pass

Audited by Gen Agent Trust Hub on Sep 21, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches framework updates from the vendor's official GitHub repository at github.com/thaitype/chief.git using git clone and git ls-remote commands.
  • [REMOTE_CODE_EXECUTION]: The skill executes a maintenance script (upgrade.sh) located within the downloaded repository content. The instructions specify running the script in both planning and application modes.
  • [COMMAND_EXECUTION]: Several shell commands are used to manage the upgrade process, including git, bash, diff, cp, ln, and rm for file manipulation and environment detection.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from a remote repository which could theoretically contain instructions targeting the agent's logic.
  • Ingestion points: External code and templates are ingested via git clone in SKILL.md.
  • Boundary markers: Explicit boundary markers for script output are absent, though the skill requires manual user review of diffs and plan outputs before proceeding.
  • Capability inventory: The skill utilizes shell execution (bash, git) and file system operations (cp, ln, rm) across the upgrade process.
  • Sanitization: There is no automated sanitization of the downloaded framework content, relying instead on user approval and author trust.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 21, 2026, 11:41 AM