Changelog

You can now pass an optional document_id when you add an episode to a standalone graph, grouping chunks of the same source — pages of a PDF, sections of a JSON export, a channel’s messages — the way threads group messages on a user graph. Episodes that share a document_id resolve pronouns and references against each other during ingestion, and Zep summarizes the document as it grows. New endpoints list the episodes belonging to a document or a thread and return a graph’s document summaries.

Connected MCP clients can now follow entity relationships instead of only searching memory. New tools return a bounded multi-hop neighborhood around seed entities, an entity’s direct neighbors with the connecting facts, and a list of raw ingested episodes; graph search adds nodes and edges scopes that return the entity UUIDs traversal needs. Search remains the first step, and traversal answers questions about how entities relate. On connections that restrict access with UserGroup policies, the standalone-graph variants of the new tools need their own grants; an existing graph.search grant does not cover them.

An Account Owner can now configure enterprise single sign-on without handing identity-provider credentials to Zep. Under Account ▸ Settings ▸ Enterprise SSO, submit your issuer URL, client ID, and client secret, and request the email domains your members sign in with. Zep approves each domain request, then validates and enables the connection; nothing changes for members until it is enabled. Available on managed cloud.

Guidance for choosing an ingestion method by cadence, separating live conversation turns from on-disk and recurring imports, was added to the Ingest section.

Read the docs

You can now invite teammates to a project as a Project Admin, without granting them account-level access. From Project ▸ Members, a Project Admin adds an existing account member or invites a new teammate with a project-scoped role. Available on accounts with role-based access control enabled.

Read the docs

Signing in to connect an MCP client now treats identity-provider issuer URLs that differ only by a trailing slash as equivalent. Some providers issue tokens with a trailing slash on the issuer, which previously caused an otherwise successful sign-in to fail with an identity verification error.

Read the docs

The Memory MCP server’s tools now expose human-readable titles and read-only and destructive hint annotations. MCP clients and directories that surface this metadata show clearer tool names and indicate which tools only read your memory versus modify it.

Read the docs

You can now walk the Context Graph directly instead of fetching capped edge lists and resolving nodes yourself. New endpoints return a node’s neighbors, a bounded multi-hop subgraph around seed nodes, and paginated episode listings for a user or graph.

• Search and list endpoints accept node- and episode-anchored filters (connected_node_uuids, source_node_uuids, target_node_uuids, episode_uuids) to select edges by their endpoints or the episode they were derived from.
• Edge responses now include the source and target node names and labels, so a second lookup is no longer needed to display an edge.
• The neighbors and subgraph endpoints traverse in any direction — incoming, outgoing, or both.

Listing nodes and edges now honors episode_metadata_filters and exact_metadata_match_filter. Previously these filters were accepted but silently ignored, so a filtered list request returned unfiltered pages. Requests that send these filters now receive filtered results.

Read the docs

Ingestion now keeps a new fact even when one of its entities was referenced only as an edge endpoint. Previously such facts could be discarded, so a later value never superseded the earlier one and search kept returning the stale fact.

Read the docs

When you bring your own OpenAI model, you can now set a custom base URL to route requests to a regional OpenAI host (for example, eu.api.openai.com) for data residency. The field is on the model provider configuration in the dashboard; leaving it empty keeps the default host.

Read the docs

Enterprise single sign-on now treats issuer URLs that differ only by a trailing slash as equivalent. Some identity providers issue tokens with a trailing slash on the issuer, which previously caused an otherwise successful login to be rejected.

A Strands integration guide for adding long-term agent memory was added to the Agent Frameworks section.

Read the docs

An Eve integration guide for adding agent memory to Vercel Eve agents with the Zep SDK was added to the Agent Frameworks section.

Read the docs

Zep now documents the zep-strands package for Strands. The guide covers ZepMemoryStore with MemoryManager, batched server-side extraction, provisioning helpers, and an optional pin-or-expose graph-search tool.

Read the docs

Project administrators can now stop an MCP connection’s clients from reaching standalone graphs while still allowing access to identity-bound user graphs. The control appears on both Google Workspace and custom OIDC connections and is on by default, so existing connections are unchanged; disable it where clients should see only their own user graph.

Read the docs

When an account is at its project limit, the dashboard now greys out every create-project control with an explanatory tooltip instead of letting you start a flow the server would reject. Previously, launching “Create with wizard” at the limit produced a confusing error partway through and left the run stranded.

Listing a graph’s episodes now returns episodes that are still processing, not only ones that finished. Newly accepted episodes previously stayed absent from the listing until processing completed.

Read the docs

The onboarding wizard no longer reports a just-created project as missing, collapses every backend failure into a generic 500, or lists shadow episodes that have no stored record. Failed sample generation now returns a clear message instead of an opaque error.

Access-policy documentation now lives in a dedicated Governance section, and attribute-based access control (ABAC) has been retitled “policy-based agent access.”

Read the docs