dify-docs-editor-test
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external documentation drafts through the
{DRAFT}placeholder. This creates a surface for indirect prompt injection where a malicious draft could attempt to influence the sub-agent's judgment or behavior. - Ingestion points: External content is loaded from the file path provided in the
{DRAFT}variable within the Dispatch prompt. - Boundary markers: The prompt uses a numbered list structure to separate instructions from the draft content, though it lacks explicit structural delimiters like XML tags or triple-backticks specifically for the draft text.
- Capability inventory: The skill description does not specify access to dangerous tools (such as network or shell access); it focuses on text analysis and reporting.
- Sanitization: The skill mitigates risks by providing explicit negative constraints to the sub-agent, such as 'Read nothing else and run no other command' and 'Do not check facts, formatting, links, or terminology'.
Audit Metadata