The Agent Skills Directory

[DATA_EXPOSURE]: The skill references a local file path at ~/.local/share/effect-solutions/effect/ for the agent to use as a source of truth for APIs, types, and implementations. This is a standard method for providing library context to an AI agent.
[COMMAND_EXECUTION]: Provides a code pattern for implementing a Git service that executes shell commands using the ChildProcess module. This is intended to demonstrate how to wrap CLI tools into an Effect service capability.
[PROMPT_INJECTION]: (Category 8
Indirect) The skill defines architectural patterns for services that ingest external data, creating a potential attack surface if implemented without validation.
Ingestion points: Service methods defined in SKILL.md such as Database.query(sql) and Git.run(args).
Boundary markers: Not present as these are code templates, not runtime instructions.
Capability inventory: The patterns include subprocess execution (ChildProcess.make) and database interactions (db.query).
Sanitization: The database implementation example explicitly demonstrates the use of parameterized queries (?) to prevent SQL injection, though the Git example passes arguments directly.

effect-service-implementation