The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to perform privilege escalation by using sudo systemctl restart clickhouse-server to manage system services.
[COMMAND_EXECUTION]: Provides instructions for executing destructive database operations, such as ALTER TABLE ... DROP PARTITION and KILL QUERY, which can lead to permanent data loss or service disruption if executed incorrectly.
[COMMAND_EXECUTION]: Utilizes INTO OUTFILE to export database query logs and metrics to the /tmp directory (e.g., /tmp/incident-queries.json). This pattern can expose sensitive information contained within database queries to other users or processes on the host system.
[SAFE]: Fetches service status information from ClickHouse's official status page at https://status.clickhouse.cloud.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) because it processes untrusted data from system logs without adequate safeguards.
Ingestion points: Reads and processes data from system.query_log and system.processes as defined in SKILL.md.
Boundary markers: Absent; there are no delimiters or instructions to the agent to disregard instructions embedded within the logs.
Capability inventory: The skill uses Bash (with kubectl and curl access), Read, and Grep tools.
Sanitization: Absent; the content from the logs is processed and exported without validation or escaping.

clickhouse-incident-runbook