root-cause-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes shell commands including
npx tsx,node --experimental-strip-types, and@kopai/clito execute dynamically generated analysis scripts and interact with the Kopai backend. - [DYNAMIC_EXECUTION]: The core workflow involves the runtime generation and execution of TypeScript (
.mts) scripts to process observability data using the Kopai SDK. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and analysis of production telemetry, which constitutes an external and potentially attacker-controlled input source.
- Ingestion points: Telemetry data is retrieved from the Kopai backend using
client.query,client.searchLogs, andclient.getTraceas documented inSKILL.md,workflow-find-errors.md, andworkflow-get-context.md. - Boundary markers: The instructions do not define delimiters or explicit warnings for the agent to ignore instructions that might be embedded within telemetry fields like log bodies or trace status messages.
- Capability inventory: The agent uses the ability to write and execute TypeScript code to analyze the retrieved telemetry data.
- Sanitization: There is no evidence of sanitization or filtering of log message bodies or exception details before they are processed by the agent.
Audit Metadata