The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the execute-sql (or posthog:execute-sql) tool to run queries against engineering_analytics_ci_failures and engineering_analytics_ci_job_history views to diagnose test failures.
[SAFE]: The skill handles potentially untrusted data from external sources such as CI error logs and commit metadata, which represents a surface for indirect prompt injection.
Ingestion points: SQL query results containing error_signature, commit_message, and test_id processed in references/investigation-queries.md.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" markers for the data returned from SQL queries.
Capability inventory: The skill is authorized to perform database queries and use specific analytical MCP tools (engineering-analytics-broken-tests, etc.).
Sanitization: There is no explicit sanitization or validation of the log content before it is presented to the agent.

investigating-ci-failures