The Agent Skills Directory

[SAFE]: The skill implements a robust planning framework for technical projects. It reads architectural and specification documents from the local project directory and generates a structured task list in a specialized artifact file. There is no evidence of unauthorized data access or exfiltration.- [SAFE]: The multi-agent architecture (decomposer, mapper, ordering, acceptance, critic) includes a built-in verification loop. The critic-agent enforces 'Critical Gates' such as ensuring every task is a vertical slice and has verifiable acceptance criteria, which serves as a safeguard against malformed or deceptive instructions.- [SAFE]: Tool usage is restricted to standard file operations (Read, Glob, Grep) and Bash for basic local tasks. No automated network connections to untrusted domains or remote code execution patterns were found. Instructions regarding network protocols (e.g., POST/GET examples) are strictly for defining user-application acceptance criteria.- [SAFE]: The skill uses a date-based freshness check on local artifacts to recommend re-runs, which is a benign maintenance feature. No persistence mechanisms or privilege escalation attempts were detected.

task-breakdown