resume-handoff
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses standard shell utilities like
git,mv,mkdir, andpwdto resolve project roots, check branch status, and manage handoff files. These actions are transparently described and limited to the local project environment.- [DATA_EXFILTRATION]: The skill reads local handoff documents stored in.claude/handoffs/. There are no network-capable commands or evidence of sensitive data being sent to external destinations.- [PROMPT_INJECTION]: Although the skill processes content from handoff files which could contain instructions, it implements a strict human-in-the-loop requirement. It explicitly forbids the agent from auto-executing steps from the handoff and requires the user to manually select the next direction, effectively mitigating indirect prompt injection risks.
Audit Metadata