The Agent Skills Directory

[SAFE]: The skill provides a comprehensive framework for debugging, focusing on root cause analysis rather than quick fixes. Instructions encourage finding the source of bugs before attempting implementations, promoting secure and stable code practices.
[COMMAND_EXECUTION]: The provided utility script find-polluter.sh and instrumentation examples in the documentation use local commands for legitimate development purposes, such as running test suites or verifying environment configurations. No dangerous command patterns, privileged operations, or external exfiltration attempts were identified.
[INDIRECT_PROMPT_INJECTION]: The skill involves the ingestion of external diagnostic data such as logs and error messages. While this presents a theoretical attack surface for indirect prompt injection, the instructions focus on structured technical analysis and do not incorporate unsafe data handling practices or instructions that would cause the agent to deviate from its safety constraints.

systematic-debugging