The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to read and implement instructions from project files like ADRs and design documents. While these files are necessary for the skill's operation, they could potentially serve as a vector for indirect prompt injection if they contain untrusted content.
Ingestion points: The agent reads project specifications from docs/adr/, docs/scope/, and design.md.
Boundary markers: There are no explicit instructions for the agent to use delimiters when processing content from these files.
Capability inventory: The skill allows for code modification and command execution via the Write, Edit, and Bash tools.
Sanitization: The instructions focus on implementation fidelity rather than sanitizing input from the specification files.
[COMMAND_EXECUTION]: The skill performs shell commands necessary for modern software development.
Evidence: The skill executes project initialization commands (e.g., create-next-app, cargo new), dependency installation, and database migration tasks as part of its core build flow.
[EXTERNAL_DOWNLOADS]: The skill fetches resources from well-known and trusted external services.
Evidence: The UI track instructions reference Google Fonts and established placeholder image services (e.g., Unsplash, Lorem Picsum) to support rapid prototyping.

develop