The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands and project scripts to perform verification, specifically using bun run lint for typechecking, bun run tf:plan:staging for infrastructure planning, and ./skills/dev-browser/server.sh for browser testing.
[CREDENTIALS_UNSAFE]: The skill is designed to access local environment variables (DEV_DEMO_EMAIL, DEV_DEMO_PASSWORD) from the .env file for automated authentication and relies on AWS credentials for executing Terraform plan operations.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it determines its verification strategy based on the contents of the scripts/ralph/prd.json file.
Ingestion points: The scripts/ralph/prd.json file, specifically the userStories array and its acceptance criteria strings.
Boundary markers: No specific delimiters or instructions to ignore instructions within the data are used during processing.
Capability inventory: The skill has the ability to execute subprocesses (via bun and bash), read and write project files, and perform browser-based visual verification.
Sanitization: The skill does not implement validation or sanitization of the strings found in the PRD before classifying them into verification strategies.

ralph-verify-acceptance-criteria