context-management
Pass
Audited by Gen Agent Trust Hub on Mar 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists of instructional text and templates for managing session state. No malicious patterns, prompt injections, or unauthorized code execution vectors were detected.
- [DATA_EXPOSURE]: The skill describes a protocol for writing task progress and environment state to a local file (
.autopilot/handoff.md). - Evidence: This activity is confined to local file system operations intended for session persistence and does not involve hardcoded credentials or transmission of data to external servers.
- [COMMAND_EXECUTION]: The skill recommends using standard CLI tools like
grepfor targeted file searching to minimize context usage. - Evidence: These tools are suggested as part of a performance optimization strategy rather than as a mechanism for executing unvalidated external input.
Audit Metadata