The Agent Skills Directory

[SAFE]: Analysis of the skill instructions and configuration revealed no indicators of malicious intent, unauthorized data access, or code obfuscation. The behavior is consistent with the stated purpose of development task management.
[DATA_EXFILTRATION]: The skill interacts with project files in the .scratch/ directory and communicates with established issue tracking platforms (GitHub, Linear). These activities are restricted to task publication and use well-known service providers for their intended purpose.
[PROMPT_INJECTION]: The skill ingests external data from specifications and issue comments, creating a surface for indirect prompt injection.
Ingestion points: Reads plan/spec files and external issue bodies/comments referenced by the user.
Boundary markers: None explicitly defined in the prompt instructions.
Capability inventory: File system reading, codebase exploration, writing to local files, and interaction with tracker APIs.
Sanitization: No explicit filtering or sanitization of external content is described.
Assessment: The vulnerability surface is managed by a mandatory user review phase where the agent must present the ticket breakdown for approval before any publication occurs.

to-tickets