The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides examples of diagnostic instrumentation for multi-component systems that involve inspecting the environment.
SKILL.md contains examples of checking signing identities with the macOS security tool and inspecting environment variables to verify configuration propagation.
The skill demonstrates safe practices for checking variable existence without revealing the value (e.g., ${IDENTITY:+SET}${IDENTITY:-UNSET}).
[COMMAND_EXECUTION]: The skill includes a utility script, find-polluter.sh, designed to identify tests that create unwanted state or side effects.
This script dynamically executes local test files matching a user-defined pattern using the npm test command.
[INDIRECT_PROMPT_INJECTION]: The debugging workflow relies on the agent processing external data sources that could potentially contain malicious instructions.
Ingestion points: The agent is instructed to analyze error messages, stack traces, and test output.
Boundary markers: No specific delimiters are mandated for wrapping logs or external evidence within the agent's context.
Capability inventory: The agent possesses standard development capabilities including shell command execution (npm, git) and filesystem access.
Sanitization: No specific filtering or sanitization of ingested logs is defined in the instructions.

systematic-debugging