The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it instructs the agent to ingest and process data from external, potentially untrusted sources like Kubernetes logs and events.\n
Ingestion points: Instructions in SKILL.md to run kubectl logs, kubectl get events, and kubectl describe pod.\n
Boundary markers: No specific delimiters or safety instructions are provided to help the agent distinguish between its instructions and the content of retrieved logs.\n
Capability inventory: The agent has the capability to execute shell commands (kubectl), read environment variables (env), and inspect file system contents via exec.\n
Sanitization: There are no explicit mechanisms mentioned for filtering or sanitizing the retrieved diagnostic data before processing.\n- [COMMAND_EXECUTION]: The skill facilitates the execution of diagnostic shell commands and local repository scripts.\n
Evidence: SKILL.md includes commands to run local scripts deployment/collect-logs.sh and deployment/diagnose.sh.\n
Evidence: Multiple instances of kubectl exec are used to run tools like redis-cli, df, du, and stat for system diagnostics.

debug-buttercup