byted-viking-aisearch-feishu
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes content from external sources such as Feishu docs, sheets, and wikis. This content is returned to the agent's context, where it could potentially contain instructions that influence the agent's behavior.
- Ingestion points: Data is ingested via Feishu Open API calls within
scripts/feishu_search.pyusing methods likeget_docx_raw_content,get_doc_raw_content,fetch_sheet_content, andfetch_bitable_content. - Boundary markers: The instructions do not define clear delimiters or "ignore instructions" wrappers for the content fetched from the documents.
- Capability inventory: The skill is restricted to network operations targeting the
open.feishu.cndomain. It does not have capabilities for local file writing, command line execution, or persistent access. - Sanitization: Content is converted to structured JSON, TSV, or plain text strings before being passed to the agent, reducing but not eliminating the risk of instruction obedience.
Audit Metadata