The Agent Skills Directory

[SAFE]: The skill consists of instructional markdown guidelines for debugging. No malicious scripts, remote dependencies, or safety bypasses were found. The methodology promotes best practices for technical investigation.\n- [COMMAND_EXECUTION]: The skill includes shell command templates for diagnostic instrumentation, such as environment variable checks and state logging. These are benign examples for the agent to adapt and do not constitute an automated execution vulnerability.\n- [INDIRECT_PROMPT_INJECTION]: The skill manages potential injection surfaces in failure data through a rigid reasoning framework. 1. Ingestion points: Bug reports and system logs provided as input. 2. Boundary markers: The four-phase methodology acts as a logical reasoning boundary. 3. Capability inventory: The agent is expected to perform file operations and command execution for debugging. 4. Sanitization: The methodology relies on verification of evidence rather than input sanitization.

systematic-debugging