decision-maker

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill manages git operations including git_commit and git_push. These actions are central to its workflow for version control and code management as described in the Git Workflow section of SKILL.md.\n- [PROMPT_INJECTION]: The agent is susceptible to indirect prompt injection through the ingestion of external data.\n
  • Ingestion points: The skill processes "decision beads" from other agents or workflows as described in its primary purpose (SKILL.md).\n
  • Boundary markers: Absent; there are no specified delimiters to isolate bead content from the agent's internal instructions or to warn the agent to ignore instructions within the beads.\n
  • Capability inventory: The agent can execute git commands (git_commit, git_push) and signal completion (done) through the provided JSON action format.\n
  • Sanitization: No sanitization logic, escaping, or validation of external input from the beads is documented in the skill instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 06:29 AM