The Agent Skills Directory

[SAFE]: The skill is a legitimate orchestrator that manages a structured development pipeline (Planning, Architecture, Development, QA, and Integration) without any suspicious activity.
[COMMAND_EXECUTION]: The skill uses benign shell commands like ls, cat, and grep to verify file existence and check task completion status within the project folders (project-specs/, project-tasks/, project-docs/). These commands are essential for the workflow logic and do not access sensitive system paths.
[PROMPT_INJECTION]: The skill ingests untrusted data from project specification files, creating an inherent surface for indirect prompt injection.
Ingestion points: Files in project-specs/ and project-tasks/ are read and processed to drive the development workflow.
Boundary markers: The skill does not use explicit boundary markers or delimiters when interpolating data from files into instructions for other agents.
Capability inventory: The skill can read local files and spawn specialized agents to perform development and testing tasks.
Sanitization: No sanitization or validation of the content within ingested project files is performed.

Agents Orchestrator