The Agent Skills Directory

[COMMAND_EXECUTION]: The skill runs various shell commands to validate the environment and build status, including pnpm build, pnpm dev, and curl against localhost. These are routine operations for continuous integration and testing workflows.
[EXTERNAL_DOWNLOADS]: The skill installs the @browserbasehq/stagehand and zod packages. Browserbase is a well-known service for browser automation, and Zod is a standard validation library; both are appropriate for the skill's stated purpose.
[PROMPT_INJECTION]: The skill processes content from local documentation files which introduces a surface for indirect prompt injection.
Ingestion points: Reads instructions and requirements from docs/tasks/<feature-name>/PRD-<feature-name>.md, docs/epics/<epic-name>/USER-JOURNEY.md, and docs/epics/<epic-name>/EPIC-<epic-name>.md to generate E2E test scripts.
Boundary markers: No explicit delimiters or system instructions are provided to the agent to distinguish between the skill's logic and potentially malicious instructions embedded in the documentation files.
Capability inventory: The skill has permissions to execute shell commands, read/write files in the local repository, and perform network requests through the Stagehand browser automation tool.
Sanitization: There is no evidence of content sanitization or validation of the ingested documentation content before it is used to influence agent actions.

tester