The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes project documentation (Keel files) and source code, which are considered untrusted data sources.
Ingestion points: The skill reads 01-requirements.md, 02-system-design.md, 03-test-cases.md, 05-test-report.md, and source files within the workspace_context.code_roots.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands within the processed data are provided.
Capability inventory: The skill utilizes Bash, Read, Glob, Grep, and potentially browser-based MCP tools (Playwright, Chrome DevTools) for automation.
Sanitization: There is no explicit sanitization or validation of the content read from the documents before it is used to drive agent behavior or command execution.
[COMMAND_EXECUTION]: The --live mode involves executing shell commands to start the application (retrieved from package.json or project documents) and using browser automation tools. This capability is driven by potentially untrusted data parsed from the project files.

verify