Unbounded effective memory
Store grows forever. Model context stays bounded. Retrieval, ranking, and consolidation decide what enters working context.
memory runtime for ai agents
Give your AI agentinfinite memory.
Infinicon is a memory SDK and reference server. Persist knowledge across sessions, recall what matters, assemble bounded context from an ever-growing store.
quick start
import { openMemory } from "@infinicon/sdk"
const memory = openMemory()
await memory.remember("user prefers dark mode")
const prior = await memory.recall("theme")What it does
not a chatbot · not a vector db · the memory layer
Memory semantics, not just vectors
Episodes, atoms, consolidations, provenance, tombstones. Real memory primitives — not raw similarity search.
SDK + reference server
npm install @infinicon/sdk. npx @infinicon/server. recall, remember, assemble-context — three lines to start.
Spec-first, production-grade
Normative v0 specs, OpenAPI contracts, ADRs. Swap storage, embedders, rankers without rewriting agent logic.
Plugin host
Storage adapters, embedders, rankers, consolidators. Extend without forking the runtime.
Framework agnostic
Not a chatbot framework. Not orchestration. The memory layer your agent plugs into.
Install in 60 seconds
agent (client)
npm install @infinicon/sdk
memory server
npx @infinicon/server
→ http://localhost:8787
Run the server separately. Point the SDK at it via INFINICON_BASE_URL. Optional Postgres via DATABASE_URL.
Memory API v0
Stable HTTP contract. Ingest episodes, query by scope, hydrate refs, assemble bounded working context for any agent task.
Read spec- POST /v0/ingest
- POST /v0/query
- POST /v0/hydrate
- POST /v0/assemble-context
- POST /v0/consolidate
- POST /v0/tombstone
documentation
Specs are the source of truth.
Vision, architecture, v0 specifications, ADRs, deployment guides — everything your team needs to build on Infinicon.
Open docsInfinicon — Give your AI agent infinite memory.
Infinicon is a production-grade memory runtime for AI agents — a TypeScript SDK plus reference HTTP server.
Agents need memory semantics, not just storage. Infinicon defines episodes, atoms, consolidations, provenance, tombstones, and bounded context assembly.
Unbounded effective memory: the store grows without a fixed app limit; the model still receives a caller-bounded working context assembled via retrieval, ranking, and consolidation.
What Infinicon is not
- Not a chatbot framework
- Not an orchestration framework
- Not a vector database
- Not a prompt library
Packages
- @infinicon/sdk: TypeScript client
- @infinicon/server: Reference memory server (Bun)
- @infinicon/core-types: Spec-aligned contracts and validators
- @infinicon/plugin-host: Plugin registration host
Memory API v0
- GET /health
- POST /v0/ingest
- POST /v0/query
- POST /v0/hydrate
- POST /v0/assemble-context
- POST /v0/tombstone
- POST /v0/consolidate
- POST /v0/subscribe
- POST /v0/get-job
- GET /v0/jobs/{jobId}