The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill relies on the agent-browser CLI, which is a project from the trusted vercel-labs GitHub organization.- [COMMAND_EXECUTION]: Uses shell commands to interact with the agent-browser tool for browser automation tasks, including navigation, capturing snapshots, and monitoring network activity.- [DATA_EXFILTRATION]: Accesses project-level configuration files (e.g., .claude/qa-loop.config.md) and captures browser console logs and network response metadata for debugging. This represents local data exposure for the purpose of developer testing rather than external exfiltration.- [PROMPT_INJECTION]: Identifies an indirect prompt injection surface through the analysis of external web content.
Ingestion points: Captures content via agent-browser console, agent-browser errors, and network response bodies.
Boundary markers: No explicit delimiters or instructions to ignore embedded prompts are present in the processing logic.
Capability inventory: The skill utilizes shell commands via the browser CLI and possesses the capability to modify project code to resolve bugs.
Sanitization: No sanitization or filtering of captured browser logs is implemented before processing.

qa-loop