Supermemory
adapter readyAgent memory behind learner-1: feed it text, pages, conversations or plain facts, and it keeps a memory per prospect, campaign or user that updates and forgets as facts change, then hands it back as a search or a ready-to-inject profile.
Capabilities only. Supermemory is not exposed as raw endpoints: memories live in the platform's Supermemory organization under a container tag per workspace; use memory.add, memory.search, memory.profile, memory.list and memory.forget.
Adapter ready. The Supermemory integration is built, unit-tested and priced. It goes live, and its capabilities switch from coming soon to live, the day the platform account opens. Until then the sandbox serves them in clearly labeled mock mode.
Capabilities served
| Capability | Description | Coverage here |
|---|---|---|
| memory.add | Give your agent a memory. Send text, a URL, a conversation or plain facts; learner-1 extracts what matters, links it to what it already knows and updates facts that changed. Scope it with a namespace ("lead:acme", "campaign:q4", a user id) so each subject keeps its own memory inside your workspace. | live |
| memory.search | Ask the memory a question in plain language and get the memories that answer it, ranked by similarity, with the facts they superseded and the ones they relate to. Search memories (fast, conversational), the source documents (chunks with citations), or both. | live |
| memory.profile | Everything the memory knows about a namespace, pre-computed by learner-1: the permanent facts, the recent ones, and a ready-to-paste context block for your prompt. Add a query to get the memories relevant to what you are about to do in the same call. | live |
| memory.list | What is stored in a namespace, newest first, with each memory's version, whether it is a permanent trait or an inference, and the documents it came from. Use it to audit what the agent believes before it acts. | live |
| memory.forget | Forget one memory by id, remove a source document (and what was extracted from it), or forget everything matching a plain-language instruction ("everything about Project Titan"). Run the instruction with dry_run first to see what would go. | live |
Provider names are used for identification only. Measured success-rate and latency data will be published here once a minimum sample size is reached, never extrapolated from single calls.