Loading ToolBench
MCP server for AgentRAM. Persistent memory for AI agents with temporal memory support. No vector database, no embeddings, set up in about a minute. 15 tools for personal, shared, and versioned agent memory.
Return the current credit balance for your account. Free to call (no credit deducted).
Create a shared memory namespace that multiple agents can read from and write to. Returns a namespace_key to share with each agent. Free to call (no credit deducted).
Permanently delete a stored memory. Cannot be undone. Costs 1 credit.
List all memories stored under an agent ID, ordered by most recently written. Expired memories are excluded. Costs 1 credit.
List all memories in a shared namespace, ordered by most recent. Costs 1 credit.
Retrieve a stored memory by agent ID and key. Returns 404 (and refunds the credit) if the memory does not exist. Costs 1 credit.
Retrieve a value from a shared namespace by key. Costs 1 credit.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 12 | 2026-07-28+ | v2 |
Case-insensitive text search across keys and values for a given agent. No embeddings or vector configuration required. Costs 1 credit.
Record a fact that changes over time (a user's current plan, status, preference, or any value tied to an agent_id and key). To deliberately replace a specific earlier fact, pass its assertion_id in `supersedes`, which links the two so the history stays traceable. Writing WITHOUT `supersedes` to a key that already has a live value is allowed and creates a contested state that is surfaced on the next get_assertion — this is intended, not an error. Costs 1 credit.
Store a value under a key for a given AI agent. Writing to an existing key updates the value (no duplicates created). Costs 1 credit.
Store a value in a shared namespace. Any agent with the namespace_key can write to it. Costs 1 credit.