resolve-pending-decisions
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for interactive decision resolution within a restricted environment. It manages files exclusively within the
docs/threads/directory and does not attempt to access sensitive system locations or credentials. - [PROMPT_INJECTION]: The skill processes external data from decision bundles which could contain instructions intended to influence the agent's behavior during the resolution process.
- Ingestion points: Files located in the
.pending-decisions/directory of the active thread. - Boundary markers: Individual points are framed using the structured
discussion-point.mdformat. - Capability inventory: File system operations (read, write, delete) within the thread directory and bounded execution of follow-up actions.
- Sanitization: No specific content sanitization is described for bundle data, but operations are interactive and require user approval.
- [COMMAND_EXECUTION]: The skill includes a 'Follow-through' phase that may perform actions based on the 'Suggested action' in a bundle. This activity is explicitly gated by user consent and limited to a single bounded execution per run.
Audit Metadata