The Agent Skills Directory

[COMMAND_EXECUTION]: The skill manages project lifecycle data by creating a local _workspace/ directory and generating various markdown files, such as 01_project_plan.md and 04_risk_plan.md, to document planning and risk management outcomes.
[EXTERNAL_DOWNLOADS]: The strategy involves delegating specialized tasks to other skills within the MoAI ecosystem, including moai-lawyer:legal-contract-review for legal oversight and moai-writer:korean-humanize for final content polishing.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted user-provided project details and incorporates them into generated planning documents. (1) Ingestion points: User project descriptions, resource constraints, and identified risks are extracted and stored in _workspace/00_input.md as described in references/project-tracker.md. (2) Boundary markers: No explicit delimiters or instructions are provided to the agent to treat user-provided project data as potentially untrusted content. (3) Capability inventory: The skill utilizes file system access to create, write, and organize files within the specified workspace directory. (4) Sanitization: There are no defined processes for sanitizing or validating the project data before it is used to generate downstream planning documents.

collab-roadmap