seo-title-optimization
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, such as page content, H1 tags, meta descriptions, and slugs, to generate optimized titles and update file frontmatter. This creates a potential attack surface where instructions embedded in the analyzed pages could influence the agent's output or file-writing actions.
- Ingestion points: Workflow steps 1 and 2 involve reading keywords, H1 tags, and general page content from files or potentially external sources.
- Boundary markers: The instructions do not specify the use of delimiters or boundary markers to separate external data from the skill's logic.
- Capability inventory: The skill possesses the capability to edit files, specifically updating the frontmatter
titlefield (Workflow step 7). - Sanitization: No sanitization, escaping, or validation of the ingested content is described in the workflow.
Audit Metadata