T05 · Unauthorized Access and Privilege Escalation
Warning
- Location
SKILL.md:129- Finding
Generated content is published to a hard-coded external Feishu destination
- Content
View full analysis
, doc_token=) ``` It also instructs the agent to grant access to the requester: ```text grant_to_requester=true ``` ### Technical Analysis The skill fixes the Feishu space and parent-node identifiers inside its instructions instead of obtaining them from the requester or the authenticated Feishu account configuration. Consequently, an agent following the workflow may use available Feishu credentials to create a document in a workspace destination that the requester did not select or approve. The identifiers are routing identifiers rather than demonstrated authentication secrets. Nevertheless, they establish an external data destination and cross an authorization boundary without explicit destination confirmation. Granting the requester document access after creation does not prove that the requester owns, administers, or approved the hard-coded parent workspace. Other members or administrators of that workspace may retain access according to its existing policy. The effective operation remains limited by the permissions of the credentials available to `lark-cli` and `feishu_doc`; the instructions do not independently bypass Feishu authentication. Exploitation therefore requires valid credentials that can write to the specified destination. ### Attack Path 1. A user invokes the skill and supplies a topic or content used to create a science article. 2. The agent generates the article, references, and review information. 3. The agent invokes ...[truncated 1197 chars]- Remediation
View remediation