category-page-optimizer
Pass
Audited by Gen Agent Trust Hub on Sep 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes several Python scripts to perform data analysis, schema generation, and report publishing. These include
scripts/project_memory.py,scripts/dfs_client.py,scripts/schema_gen.py, andscripts/report_publish.py. These scripts are part of the skill's internal logic and are used for its primary purpose. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites when it fetches category URLs to audit H1 tags, titles, and canonical link structures.
- Ingestion points: The skill performs web fetches on the user-provided category URL, including paginated versions and filtered variants, to inspect markup.
- Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are documented for the ingested HTML content.
- Capability inventory: The agent has the capability to execute the local Python scripts mentioned above and write report files via
report_publish.py. - Sanitization: The skill description does not specify sanitization or filtering methods for the fetched external content.
Audit Metadata