investigating-error-issue
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data originating from external applications, which could contain malicious instructions designed to influence the agent's behavior during analysis.
- Ingestion points: The skill fetches data through
posthog:query-error-tracking-issue-events(which includes stack traces and exception messages),posthog:query-logs(raw log entries), andposthog:execute-sql(event properties). - Boundary markers: None identified. The skill does not instruct the agent to use specific delimiters or to treat the content of exceptions and logs as potentially adversarial text.
- Capability inventory: The skill utilizes
posthog:execute-sqlfor database analysis andposthog:query-logsfor log correlation. - Sanitization: No explicit sanitization or filtering of the ingested content is described in the workflow.
- [DYNAMIC_EXECUTION]: The skill dynamically generates and executes SQL queries based on data points (like issue IDs and session IDs) retrieved at runtime.
- Evidence: The workflow (Steps 3, 4, and 5) provides SQL templates for tools like
posthog:execute-sql. The agent is instructed to populate these templates with parameters found in previous steps to perform breakdowns and session reconstruction.
Audit Metadata