The Agent Skills Directory

[SAFE]: The skill implements a robust research pipeline with mandatory planning, reflection, and verification phases to ensure accuracy.
[SAFE]: Phase 0 includes a local-first check using rg to search the current directory, which is a standard procedure for context-aware research.
[SAFE]: The skill explicitly guides the agent to avoid common AI pitfalls such as sycophancy, anchoring bias, and fabrication of citations.
[SAFE]: No evidence of data exfiltration, credential theft, or unauthorized remote code execution was found.

deep-research