The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local bash script located at .specify/scripts/bash/check-prerequisites.sh from the repository root. This allows for arbitrary code execution if the repository being processed is malicious.\n- [DYNAMIC_EXECUTION]: The skill implements an extension hooks system that reads .specify/extensions.yml and dynamically triggers additional commands (e.g., EXECUTE_COMMAND: {command}). This allows project configuration to control the agent's workflow and execute other skills or tools.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data to drive high-capability actions like command execution and GitHub issue creation without strict boundaries or sanitization.\n
Ingestion points: Reads and follows instructions from .specify/extensions.yml, tasks.md, and the output of project scripts.\n
Boundary markers: None identified to separate project data from agent instructions.\n
Capability inventory: Access to shell execution, dynamic command triggering, and GitHub MCP tools for reading/writing repository data.\n
Sanitization: No validation or escaping is performed on hook command names or task descriptions before execution or creation.

speckit-taskstoissues