The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes vendor-specific command-line interfaces, ironbee-browser-devtools-cli and ironbee-node-devtools-cli, to perform browser automation and connect to Node.js processes for debugging.
[DATA_EXFILTRATION]: The skill accesses network request headers, console logs, and screenshots from target websites, which may contain sensitive information such as session tokens or internal system details.
[PROMPT_INJECTION]: The skill processes content from external websites (e.g., console messages and HTTP responses) that could contain malicious instructions designed to influence the agent's behavior.
Ingestion points: Output from get-console-messages and get-http-requests in SKILL.md.
Boundary markers: Absent; no markers or warnings are used to distinguish external data from system instructions.
Capability inventory: Interaction with browser sessions and backend process debugging via ironbee-browser-devtools-cli and ironbee-node-devtools-cli.
Sanitization: No sanitization or validation of the retrieved external content is performed before presentation to the agent.

performance-audit