keyword-research
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources including Semrush MCP reports, Google Search Console performance data, user-provided exports, and manual SERP inspections. This creates a surface where malicious instructions could be embedded in the retrieved data, such as within keyword strings, page titles, or search snippets.
- Ingestion points: Data is brought into the agent context through MCP tools like keyword_research, organic_research, and competitors_research, as well as via file imports from Google Search Console and Semrush UI exports.
- Boundary markers: The instructions explicitly direct the agent to preserve raw report rows separately from the analyzed shortlist and to distinguish model judgments from source data, which helps mitigate accidental instruction following.
- Capability inventory: The skill uses MCP tools for data retrieval and performs logical analysis, clustering, and prioritization. It does not use untrusted data in sensitive operations like shell command execution or arbitrary file writes.
- Sanitization: The skill includes instructions to screen for search operators, broken encoding, and malformed tokens to clean the input data before processing.
Audit Metadata