developer-keyword-research
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted data which could contain malicious instructions meant to influence the agent's behavior.
- Ingestion points: The skill ingests support-ticket subjects, issue-tracker questions, community threads, and docs-site search logs (SKILL.md, Workflow Step 3).
- Boundary markers: The instructions lack specific boundary markers or warnings to the agent to disregard instructions embedded within the processed data.
- Capability inventory: The skill maintains task state in local files (
devrel-context.md) and is designed to interact with Search Console APIs and user-provided CSV exports. - Sanitization: The skill performs normalization on strings (Workflow Step 4) to remove specific IDs and paths, but this is intended for data clustering and PII removal rather than security sanitization against prompt injection.
- [DATA_EXPOSURE]: The skill requests the retrieval of sensitive business information for analysis.
- Evidence: The interview section (Question 3) and Workflow Step 3 instruct the agent to gather first-party search-query reports, docs-site search logs, support tickets, and product telemetry. While this is necessary for the skill's primary function, it involves the agent handling potentially confidential business data.
Audit Metadata