managing-path-cleaning-rules

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted URL path strings retrieved from database event logs to identify patterns and suggest configuration changes.
  • Ingestion points: The execute-sql tool is used to query $pathname and $entry_pathname from the events table as described in the "Inspect the real paths" section.
  • Boundary markers: Absent; there are no specific instructions or delimiters defined to isolate the data returned by the database from the agent's core instructions.
  • Capability inventory: The skill has read access via execute-sql and write access to project configuration via the path-cleaning-rules-update and project-settings-update tools.
  • Sanitization: Absent; the skill does not specify validation or sanitization routines for the sampled paths before they are used to draft regex rules.
  • [EXTERNAL_DOWNLOADS]: The skill references the official syntax documentation for the RE2 regex engine hosted on Google's GitHub repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 03:19 PM
Security Audit — agent-trust-hub — managing-path-cleaning-rules