Zep for AI agents
Zep is a memory layer for AI agents, storing facts, sessions, and context graphs, used by developers building agents that need to remember past interactions. Connect it once through Arc0, and your agent, or Claude, ChatGPT and Cursor, can use it through one MCP endpoint, limited to what each user approved.
What agents do in Zep.
Get a thread's messages
Pull the conversation history for a thread, a read-only context fetch.
Add a fact to the graph
Record a new fact tied to a user or session for future reference.
Confirm before deleting a user's memory
Check with the user before deleting a user's stored session memory, since that history can't be recovered.
41 Zep actions, graded by risk.
Every Zep action is tagged read, write or destructive, so one policy covers the whole app and new actions inherit the right default.
read
21Look things up. Allowed by default.
- zep.get_edgeGet Edge by UUID
- zep.get_userGet User by ID
- zep.get_groupGet Group by ID
- zep.get_sessionGet Session by ID
- zep.list_threadsList Threads
- zep.get_user_nodeGet User Node
- zep.get_node_edgesGet Node Entity Edges
- zep.get_user_nodesGet User Nodes
- zep.get_graph_by_idGet Graph by ID
- zep.get_task_statusGet Task Status
- zep.get_project_infoGet Project Info
- zep.get_user_threadsGet User Threads
- zep.get_user_sessionsGet User Sessions
- zep.get_session_memoryGet Session Memory
- zep.list_users_orderedList Users Ordered
- zep.get_session_messageGet Session Message by UUID
write
15Create and change things. Allow, or ask the user first.
- zep.create_userCreate User
- zep.update_userUpdate User
- zep.create_graphCreate Graph
- zep.create_groupCreate Group
- zep.update_graphUpdate Graph
- zep.update_groupUpdate Group
- zep.create_threadCreate Thread
- zep.create_sessionCreate Session
- zep.update_messageUpdate Message
- zep.update_sessionUpdate Session Metadata
- zep.add_fact_tripleAdd Fact Triple
- zep.add_session_memoryAdd Session Memory
- zep.add_thread_messagesAdd Thread Messages
- zep.clone_graphClone Graph
- zep.graph_searchGraph Search
destructive
5Delete, cancel or archive. Ask first, or deny outright.
- zep.delete_userDelete User
- zep.delete_graphDelete Graph
- zep.delete_groupDelete Group
- zep.delete_threadDelete Thread
- zep.delete_sessionDelete Session Memory
Zep in three steps.
- 01Your users connect ZepThey add their Zep api key on Arc0 Connect, under your brand. It goes straight into the vault.
- 02You set the rulesReads run, writes like “create User” can wait for the user, and “delete User” can be denied outright.
- 03Any agent can actYour agent calls Zep through the Arc0 SDK or MCP, and so can Claude, ChatGPT and Cursor. Every call lands on the audit log.
await arc0.policies.set('zep', { read: 'allow', write: 'ask', // create_user destructive: 'deny', // delete_user }) # Claude Code: the same connection, one URL $ claude mcp add --transport http arc0 \ https://mcp.arc0.ai/u/u_8f2
How Zep connects.
Users add their Zep api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
The same Zep connection serves your agent over MCP and your own backend over REST and the proxy, so a user connects once. How Arc0 handles credentials →
- AUTH
- API key
- CREDENTIALS
- Per-tenant encrypted vault
- MODEL SEES
- Results only, never credentials
- AUDIT LOG
- Every call, on every plan
Use Zep from any agent.
Zep and Arc0, answered.
Can I use Zep with Claude, ChatGPT or Cursor?
Yes. Connect Zep to Arc0 once, then add your Arc0 MCP URL to Claude, ChatGPT, Cursor, Claude Code or any other remote-MCP client. Each assistant only gets the Zep actions you allow.
How do users connect Zep?
Users add their Zep api key on Arc0 Connect. It is encrypted in the vault, never shown to the model, and each user can rotate or revoke it at any time.
Which Zep actions can my agent take?
41 in total: 21 read, 15 write and 5 destructive, such as “create User”. Your policies decide which of them each agent may call.
Can I stop my agent from deleting things in Zep?
Yes. Actions like “delete User” are graded destructive. Set destructive actions to deny, or to ask so the user approves each one, and blocked calls still show up on the audit log.
Can my own backend call Zep too?
Yes. The same Zep connection is available over REST and through the Arc0 proxy, so your product and your agent share one connection per user.
Plug Zep into your agent.
Your users connect Zep once, under your brand. Your agent gets 41 actions behind your policies, with every call on the record.
Free to build · MCP + REST · Audit log on every plan