The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The RemoteDebugServer in resources/implementation-playbook.md implements a WebSocket-based remote evaluation engine. It uses the Node.js inspector module to execute arbitrary JavaScript expressions provided by a client via the evaluate command.
[COMMAND_EXECUTION]: The skill configures the Node.js debugger to listen on all network interfaces (0.0.0.0:9229). This pattern, found in both the RemoteDebugServer class and the Dockerfile configuration, exposes the application to remote debugging attacks if network access is not restricted.
[DATA_EXFILTRATION]: The OpenTelemetry implementation in tracing.js captures and stores the full content of HTTP request bodies in spans (span.setAttribute('http.request.body', JSON.stringify(request.body));). This practice can inadvertently record and export sensitive information such as credentials, API keys, or personal user data to external tracing backends.
[DATA_EXFILTRATION]: The ProductionDebugger middleware in production-debug.js allows for the exfiltration of internal application logs and debug context via the X-Debug-Info response header when a specific request token is provided.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. Untrusted data enters the agent context via the $ARGUMENTS placeholder in SKILL.md. There are no boundary markers or sanitization logic present to delimit this input. The skill possesses extensive capabilities, including file writing in performance-profiler.js, network server creation in remote-debug-server.js, and arbitrary code execution through the inspector module.

distributed-debugging-debug-trace