The Agent Skills Directory

[COMMAND_EXECUTION]: Orchestrates development workflows by executing system commands.
Evidence: Uses git worktree add to create isolated environments and npm run for quality gates (SKILL.md).
[COMMAND_EXECUTION]: Integrates with Supabase for database operations.
Evidence: Executes supabase db push, supabase login, and supabase link (references/supabase-deployment.md).
[EXTERNAL_DOWNLOADS]: References external network requests for system verification.
Evidence: Includes a curl command to a placeholder health check endpoint at https://api.example.com/health (references/supabase-deployment.md).
[PROMPT_INJECTION]: Ingests potentially untrusted data from the local file system which represents an indirect prompt injection surface.
Ingestion points: Reads plan files from ~/.claude/plans/ for project initialization (SKILL.md).
Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore embedded commands in plans.
Capability inventory: High-privilege actions including git file manipulation, supabase database deployment, and subagent dispatch via the Agent tool.
Sanitization: Absent; no validation or escaping of plan content is described before processing.

swarm-protocol