Liminary

The materials explicitly state that no keys or environment variables are required. There is no indication that users must provide API keys, OAuth tokens, or other sensitive credentials, so credential exposure appears limited based on the available facts.

The tool connects to the declared endpoint api.liminary.io. Given the stated use case of recall across documents, notes, and meetings, some user content or retrieval queries may reasonably be sent to the service for processing; however, the endpoint is known and related to the stated function, so this alone does not justify a high-risk rating.

The system checks indicate this tool has executes-code capability, meaning it may start local processes or execute code on the host side. This is a common MCP capability, but the current materials do not document execution boundaries, accessible system capabilities, or sandboxing, so it warrants caution.

The description claims memory and recall across 'documents, notes, and meetings,' indicating expected access to user documents, notes, or meeting-related data. The materials do not specify read/write scope, whether it only indexes data, whether it persists content, or whether it follows least-privilege design, so the actual access scope should be treated with caution.

Positive signals include distribution through an official registry and updates within the last year. However, there is no open-source repository, no declared license, and very low community adoption (0 stars), which limits external auditability and dependency transparency. Overall, the source is not obviously suspicious, but auditability is limited, so caution is appropriate.