suppressing-noisy-errors
Warn
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
posthog:execute-sqltool to run dynamically generated HogQL/SQL queries. These queries are constructed by the agent at runtime to verify suppression filters against historical event data before final rule creation. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting and acting upon untrusted data from error logs. It directs the agent to pull exception messages and stack traces, which are then interpolated into property filters and SQL statements.
- Ingestion points: Error messages, exception types, and stack trace strings fetched via
posthog:query-error-tracking-issue-eventsas seen in Step 2 and Step 5. - Boundary markers: None. There are no instructions for the agent to use specific delimiters or escape sequences when placing untrusted strings into generated queries.
- Capability inventory: The skill has access to
posthog:execute-sqlfor database access andposthog:error-tracking-suppression-rules-createfor permanent data ingestion modification. - Sanitization: The skill relies entirely on manual inspection by the agent or user to identify 'suspicious' results rather than providing programmatic sanitization or validation logic for the input strings.
Audit Metadata