The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute arbitrary commands extracted from project-specific test.md files. This is the intended behavior for a testing utility but allows for code execution driven by file content.
[INDIRECT_PROMPT_INJECTION]: The skill interprets and executes instructions from openspec/changes/{change-id}/test.md without significant security validation of the commands themselves.
Ingestion points: Verification strategies are read from test.md files within the project repository.
Boundary markers: The instructions state to follow the strategy "literally" and lack explicit delimiters or safety warnings for the ingested content.
Capability inventory: The skill has access to Bash (command execution), Edit (file modification), Read (file access), and Glob (filesystem discovery).
Sanitization: It implements a masking policy for API keys (e.g., ANTHROPIC_API_KEY) within generated logs but does not filter or sanitize the commands being executed.

openspec-test