The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool extensively for automating common development tasks like repository initialization, running test suites, and managing the file system. These actions are aligned with the skill's primary function as a development assistant.
[INDIRECT_PROMPT_INJECTION]: The system processes information from external sources (web research) and local files (codebase mapping and user requirements), creating a surface for indirect prompt injection.
Ingestion points: Input from research/SUMMARY.md and PROJECT.md is used to generate implementation plans.
Capability inventory: The agents handling this data have access to tools like Bash, Write, and Edit.
Sanitization: Safety is managed through verification agents (gsd-verifier, gsd-plan-checker) that validate the output and structure of the generated work rather than strictly filtering input strings.
[DYNAMIC_EXECUTION]: The gsd-executor agent generates code and immediately executes it (e.g., via npm test) as part of its development and verification lifecycle. This is a core feature of the tool and is used to automate the implementation of user requirements.

gsd