The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a project-local bash script at .specify/scripts/bash/setup-tasks.sh to retrieve project metadata and configuration.
[DYNAMIC_EXECUTION]: The skill implements a 'hooks' framework that reads the .specify/extensions.yml file and instructs the agent to execute commands defined within it using the EXECUTE_COMMAND directive. This allows the project's local configuration to dynamically dictate which commands the agent runs during execution.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes various untrusted files from the repository to drive its logic.
Ingestion points: Project documents such as plan.md, spec.md, data-model.md, contracts/, research.md, and the configuration file .specify/extensions.yml are read and parsed by the agent.
Boundary markers: The instructions do not define strict boundary markers or 'ignore' directives to separate the ingested document content from the agent's core instructions.
Capability inventory: The skill possesses the capability to execute shell scripts and arbitrary agent commands (via hooks).
Sanitization: While the skill mentions escaping single quotes in arguments, there is no evidence of sanitization or safety filtering for the actual content extracted from the design documents or hook definitions.

speckit-tasks