The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands including pnpm dev, pnpm fraimz, and pnpm evals to manage the testing lifecycle. It also calls local repository scripts like .devcontainer/test-on-daytona.sh.
[REMOTE_CODE_EXECUTION]: The skill utilizes the Chrome DevTools Protocol (CDP) to drive and inspect applications over network ports, which constitutes a form of remote interaction and control. Additionally, it scaffolds new test code based on content from markdown files.
[DATA_EXFILTRATION]: Automated commands (pnpm fraimz --pr) are used to transmit test reports and HTML artifacts from the local environment to GitHub Pull Request comments.
[DYNAMIC_EXECUTION]: The workflow relies on ctx.eval to run dynamic JavaScript within the browser context of the application under test.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection:
Ingestion points: Reads narration and flow definitions from markdown voiceover scripts (evals/voiceovers/<id>.md).
Boundary markers: No explicit boundary markers or 'ignore' instructions are defined for the scaffolding process.
Capability inventory: The skill has access to shell execution (pnpm), code generation (scaffold), and browser-level execution (ctx.eval).
Sanitization: No validation or sanitization is described for the content of the voiceover scripts before they are used to generate code.

fraimz