The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents and provides examples for numerous command-line tools used in Node.js debugging and profiling, including node --inspect, clinic, 0x, autocannon, and security. These are standard tools used for performance analysis and system state inspection during debugging.
[DATA_EXPOSURE]: Phase 1 instructions (SKILL.md) include examples of environment variable inspection (e.g., env | grep IDENTITY). This is presented as a method to verify configuration propagation across build/deployment layers. While this is a standard debugging technique, it relies on the agent's internal safety filters to prevent unintentional logging of secrets if the environment contains sensitive keys.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data such as runtime error messages and application logs.
Ingestion points: Error messages and stack traces (SKILL.md, Phase 1).
Boundary markers: Not explicitly defined in the prompts.
Capability inventory: The skill has access to Bash, Read, Write, and Edit tools.
Sanitization: No explicit sanitization of error logs is described. The risk is considered low as the data is analyzed for debugging purposes within a controlled environment.

debugging