The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill is designed to perform document generation and structured interviewing for workflow design.\n- [DATA_EXFILTRATION]: The skill accesses local markdown files (NOTES.md and workflows/*.md) to store user routines. It does not attempt to access sensitive system directories, credentials, or perform network requests.\n- [PROMPT_INJECTION]: The skill ingests data from local workspace files. Ingestion points: NOTES.md and workflows/*.md. Boundary markers: None present. Capability inventory: Limited to reading and writing local workspace markdown files. Sanitization: None. Although there is an ingestion surface for untrusted data, the lack of dangerous capabilities (network, execution, privilege escalation) makes the impact negligible.

loop-me