The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill uses 'dotnet build' and 'nx build' to verify the generated code. These are standard development tools and their use is restricted to the local project context.
[PROMPT_INJECTION] (SAFE): The skill processes user-defined requirements through '$ARGUMENTS'. While this provides an ingestion surface for potential indirect prompt injection, the skill implements a mandatory 'Wait for Approval' step (Step 5), requiring human review of the plan before any files are created or commands are executed. 1. Ingestion points: $ARGUMENTS in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: File creation across various project directories and command execution via dotnet and nx. 4. Sanitization: Absent, mitigated by the human-in-the-loop approval process.

create-feature