The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides several shell-based diagnostic and management tools, including a triage script (clay-triage.sh) and various kubectl commands for log retrieval, deployment restarts, and environment variable configuration.
[CREDENTIALS_UNSAFE]: The runbook includes a remediation step to verify and decode API keys stored in Kubernetes secrets using kubectl get secret clay-secrets -o jsonpath='{.data.api-key}' | base64 -d. While sensitive, this is a documented administrative procedure for addressing authentication failures.
[INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by processing external data.
Ingestion points: The skill reads and evaluates the output of curl requests to the Clay platform status page and API endpoints, as well as internal application health checks.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the processing steps.
Capability inventory: The agent is granted shell execution (bash), network access (curl), and infrastructure management capabilities (kubectl).
Sanitization: No sanitization or validation of the remote service output is performed before it is presented to the agent context.
[EXTERNAL_DOWNLOADS]: The skill performs connectivity tests and status checks against clay.com, api.clay.com, and status.clay.com. These represent official endpoints for the SaaS service managed by this runbook.

clay-incident-runbook