The Agent Skills Directory

[SAFE]: No malicious patterns or high-risk behaviors were detected. The skill's operations are limited to reading local log artifacts and writing consolidated markdown files within the repository's 'history/learnings/' directory.\n- [SAFE]: The skill implements strong defensive measures against prompt injection. It specifically instructs the agent to treat all '.codex' artifact content as untrusted evidence rather than instructions, and it explicitly forbids following any behavioral directives found within that data.\n- [SAFE]: Mandatory data sanitization is built into the workflow. The skill requires the agent to redact secrets and Personal Identifiable Information (PII) from extracted evidence during the 'Phase 3: Extract Durable Candidates' process and before any summary output.\n- [SAFE]: The skill adheres to the principle of least privilege, requiring explicit user approval before modifying sensitive files like 'critical-patterns.md' and providing clear options for handling ambiguous data matches instead of making silent decisions.

dream