The Agent Skills Directory

[SAFE]: The skill performs file read and write operations within the .oh/ directory to manage session persistence. This is used to maintain planning context across interactions and follows standard local storage practices for this environment.\n- [SAFE]: Integration with platform-specific tools like RNA MCP and Open Horizons MCP allows the skill to retrieve repository context and tribal knowledge to enrich the planning process. These are vendor-sanctioned integrations used as intended.\n- [SAFE]: The skill implements a task-specific logic for mapping constraints and identifying X-Y problems without utilizing risky behaviors such as remote code execution, network exfiltration, or unauthorized privilege escalation.\n- [SAFE]: While the skill ingests untrusted data from the repository (via file reads and search results), its capabilities are limited to producing text-based problem maps, posing a negligible risk of indirect prompt injection. Boundary markers and sanitization are not explicitly defined, but the intended usage remains within safe functional limits.

problem-space