The Agent Skills Directory

[PROMPT_INJECTION]: No override instructions, bypass markers, or jailbreak patterns were detected in the skill instructions or code snippets. The skill implements legitimate meta-cognitive learning processes intended for agent optimization.- [DATA_EXFILTRATION]: No evidence of unauthorized data access or credential theft was found. The skill utilizes a local database file (./reasoning-bank.db) for persistence, which is a standard and safe practice for the described functionality.- [EXTERNAL_DOWNLOADS]: The skill requires prerequisites agentic-flow (v1.5.11+) and AgentDB (v1.0.4+). These are identified as vendor-owned resources belonging to the agenticsorg ecosystem and are used for core infrastructure.- [SAFE]: No obfuscation, dynamic code generation, or privilege escalation techniques were identified within the skill's logic.

ReasoningBank Intelligence