The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell scripts (validate-entities.sh, cascade-rename.sh) and a Python script (promote-shadow.py) for administrative tasks such as data validation, refactoring, and file management within the project directory.
[EXTERNAL_DOWNLOADS]: Leverages the Agent tool to conduct web research and analyze external documentation, which is necessary for gathering competitive intelligence and technical feature details.
[PROMPT_INJECTION]: Ingests untrusted data from external websites, user-provided documents, and local codebases during 'Deep Dive' research tasks, creating an indirect prompt injection surface. Evidence: 1. Ingestion points: External URLs and local source code analyzed via Agent tool. 2. Boundary markers: Not explicitly specified in the prompts. 3. Capability inventory: Access to Bash, Write, Edit, and Agent tools. 4. Sanitization: No explicit content filtering mentioned.
[COMMAND_EXECUTION]: Utilizes dynamic shell command substitution (e.g., $(ls -td ... | head -1)) to automatically resolve its own installation path if the environment variables are not pre-configured.

features