context-handover

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's primary function is to preserve session state between context window refreshes. It performs expected file operations within the project repository to maintain continuity.
  • [COMMAND_EXECUTION]: The skill automates the creation and modification of state files in the .harness/ and .claude/ directories. These operations are restricted to the local file system and are used solely for tracking project progress.
  • [DATA_EXFILTRATION]: Progress summaries are posted to GitHub issues if the environment is configured with a GitHub remote and issue tracker metadata. This behavior is documented, user-initiated via the skill trigger, and limited to project-related updates.
  • [PROMPT_INJECTION]: The skill uses GitHub issue labels, titles, and body keywords to automatically detect the current project phase (e.g., discuss, plan, execute). 1. Ingestion points: GitHub issue metadata (SKILL.md). 2. Boundary markers: None specified for the external issue content. 3. Capability inventory: Writing session state files and posting GitHub comments. 4. Sanitization: Not specified for the ingested phase-detection data. This ingestion is used for state classification and does not influence core agent instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 10:13 PM
Security Audit — agent-trust-hub — context-handover