The Agent Skills Directory

[SAFE]: The skill's purpose is to implement cognitive architectures for multi-agent problem solving. The instructions and templates provided in the reference and setup files are legitimate educational and functional materials for complex reasoning workflows.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes user-provided problem statements through multiple sub-agents.
Ingestion points: User-supplied problem descriptions provided to the orchestrator task.
Boundary markers: The skill uses strict role-specific templates and 'information barriers' (defined in references/) to isolate agent contexts and prevent cross-contamination of reasoning.
Capability inventory: The skill uses the platform's Task system for sub-agent orchestration; the orchestrator setup includes Google Search connectivity for information gathering.
Sanitization: The skill relies on structural isolation between agents (e.g., the 'Architect' role is denied access to current implementation details) to mitigate bias and injection risks.

thinking-frameworks