The Agent Skills Directory

[COMMAND_EXECUTION]: The skill implements a hook system that reads commands from a project file (.specify/extensions.yml) and instructs the agent to execute them using the EXECUTE_COMMAND: {command} pattern. This occurs in both the Pre-Execution and Post-Execution phases.
[DYNAMIC_EXECUTION]: The skill dynamically identifies and invokes hooks based on the contents of the project's YAML configuration. It specifically requires the agent to execute mandatory hooks and wait for their result before proceeding with the main task.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through project configuration files which can lead to unauthorized actions.
Ingestion points: The skill reads instructions and commands from .specify/extensions.yml and project preferences from .specify/init-options.json and .specify/feature.json.
Boundary markers: There are no boundary markers or security warnings present to prevent the agent from executing malicious commands found in these files.
Capability inventory: The agent is granted capabilities to execute arbitrary shell commands, create directories (mkdir -p), and write files to the file system.
Sanitization: No sanitization or validation of the hook commands is performed before execution, allowing any string defined in the configuration to be executed as a command.

speckit-specify