The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool restricted to the openspec namespace. It executes specific commands such as openspec list, openspec status, and openspec instructions to retrieve metadata about the project state. This is a controlled use of command execution aligned with the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, user-provided markdown files (tasks, specs, design) to verify implementation requirements. These files represent untrusted data that could be crafted to mislead the agent's logic during the verification process.
Ingestion points: Reads content from file paths provided by the openspec instructions apply command, specifically within contextFiles.tasks, contextFiles.specs, and contextFiles.design (SKILL.md).
Boundary markers: The instructions do not define explicit boundary markers or protective delimiters to separate ingested content from the system prompt.
Capability inventory: The agent possesses the capability to execute openspec CLI commands and read/write files within the repository context.
Sanitization: The skill relies on basic markdown header parsing (e.g., '### Requirement:') to extract data, but it does not implement formal sanitization or validation of the text within those sections.

openspec-verify-change