The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes architecture for "agent-driven first response" workflows where diagnostic AI agents (such as Checkly Rocky or Honeycomb Canvas) ingest production telemetry and runbooks to identify root causes.
Ingestion points: Production failure telemetry, error messages, and per-probe runbooks defined in references/platforms-and-ci.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are provided in the implementation examples for data ingested by diagnostic agents.
Capability inventory: Diagnostic agents described in SKILL.md have the capability to analyze multi-source telemetry and provide root-cause candidates to users.
Sanitization: The implementation patterns do not include explicit sanitization of external telemetry data before it is processed by the diagnostic AI agents.
[COMMAND_EXECUTION]: The skill provides configuration templates for automated execution of monitoring scripts.
Evidence: references/platforms-and-ci.md contains a GitHub Actions workflow that executes npx playwright test and a reporting script via node scripts/report-synthetic-results.js. These are standard commands for the intended synthetic monitoring purpose.
[EXTERNAL_DOWNLOADS]: The skill references and downloads official automation tools and libraries.
Evidence: The GitHub Actions workflow in references/platforms-and-ci.md uses official actions (actions/checkout, actions/setup-node) and installs the Playwright browser automation framework (npx playwright install). These are standard dependencies from trusted sources.

synthetic-monitoring