The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to read .specify/extensions.yml and automatically execute commands registered under hooks.before_specify and hooks.after_specify. Mandatory hooks (where optional: false) are executed by the agent using the EXECUTE_COMMAND pattern without a separate confirmation step from the user.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill processes external data from project files including .specify/extensions.yml, .specify/init-options.json, and .specify/memory/constitution.md.
Boundary markers: The instructions do not define boundary markers or safety delimiters for the data read from these configuration files.
Capability inventory: Across its execution flow, the skill can create directories, write specification files, write JSON metadata files, and invoke other agent commands via the hook system.
Sanitization: There is no evidence of sanitization or validation for the hook commands or project principles loaded from the filesystem.
[DYNAMIC_EXECUTION]: The skill dynamically generates command identifiers at runtime by transforming strings found in the YAML configuration (e.g., converting dot-notation identifiers like speckit.git.commit into command paths like /speckit-git-commit), which are then directly invoked.

speckit-specify