The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides the scripts/layered-isolation-probe.sh Bash script, which uses curl and ssh to execute network probes. It uses eval to process dynamically constructed command strings based on user-provided variables like HOST and SERVER_SSH. These capabilities are essential for the skill's intended purpose of infrastructure diagnostics.- [SAFE]: No evidence of malicious obfuscation, data exfiltration, or unauthorized remote code execution was detected. The skill provides legitimate diagnostic tools, such as a Flask-based mock upstream server (scripts/mock-idle-upstream.py), which are used for controlled testing and follow industry best practices.- [SAFE]: The skill exhibits high transparency, including a detailed case study and documentation of common cognitive biases in troubleshooting. This instructional approach encourages users and agents to follow rigorous, evidence-based procedures rather than making assumptions.- [SAFE]: Indirect Prompt Injection surface analysis: The skill ingests untrusted data from user-reported error logs and environment descriptions (SKILL.md Step 0). Boundaries are maintained via a structured workflow that mandates verifying premises (Step 0.5) and a counter-review process (Step 6). Capabilities include shell and SSH access via the provided diagnostic scripts. Sanitization is achieved through the agent-driven verification and human-in-the-loop oversight emphasized in the methodology.

debugging-network-issues