The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local PowerShell script located at .specify/scripts/powershell/resolve-template.ps1 to resolve templates during its execution flow.
[DYNAMIC_EXECUTION]: The skill implements a hook system that reads command names from .specify/extensions.yml, transforms them (replacing dots with hyphens), and executes them using the EXECUTE_COMMAND directive. This allows project configuration files to trigger the execution of other agent commands.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from multiple sources including user-provided $ARGUMENTS, existing project documentation, and the .specify/extensions.yml configuration file.
Ingestion points: User input via $ARGUMENTS, project files like README.md, and the extension configuration file .specify/extensions.yml.
Boundary markers: The skill contains a 'Scope Guard' section that instructs the agent to classify input and explicitly prohibits executing feature implementations or code generation requests found in the input.
Capability inventory: The skill has the ability to execute shell scripts (PowerShell) and invoke other agent commands via the hook system.
Sanitization: The skill includes instructions to ask for clarification if instructions are unclear and to defer non-governance intents to a 'Next Actions' section rather than executing them immediately.

speckit-constitution