dbk-pickup
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use read-only Git commands (such as
git status -sandgit diff) to verify the state of the workspace against the handoff report. This is a standard practice for context alignment and does not involve high-risk commands. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external files (handoff reports in
.agent/handoff/). While this represents a data ingestion surface, the skill implements strong mitigation strategies: it mandates a validation of thehandoff_version, focuses only on specific core sections of the report to avoid context bloat, and explicitly prohibits the blind reading of raw session transcripts. Furthermore, the agent is instructed to align with the user and wait for confirmation before performing modifications if the workspace state has changed. - [SAFE]: The skill provides a structured workflow for task resumption with an emphasis on verification and user alignment. The vendor resource prefix 'dbk-' aligns with the author 'DBinK'.
Audit Metadata