Features
The full stack for production-grade agent memory infrastructure — graph backends, retrieval, observability, and the controls a security review asks for. Eleven entries; status is on every cell.
Everything in the memory layer.
Shipped capabilities first; roadmap entries are marked on the cell so nothing reads as further along than it is.
Multi-Graph Backends
Seamlessly switch between PostgreSQL, FalkorDB, SurrealDB, and more. Your agent's memory isn't locked into any single vendor.
Graph-Based Memory
Store agent interactions as a knowledge graph — entities, relationships, and facts that your agents can traverse and reason over.
File & Blob Attachments
Attach files to sessions as first-class, S3-backed objects — preserved, not just digested to text. Messages reference them by blob_id, and their content is extracted into the fact store. Memory that keeps the file.
Temporal Memory & Fact Supersession
Corrections supersede instead of delete. Read facts at any point in time — as-of-now and as-of-past queries — with observable supersession events (superseded_count, webhooks) and edges that expire in lockstep.
Multi-LLM Support
Works with OpenAI, Anthropic, Azure, OpenRouter, and local models via Ollama. Route tasks to different providers based on cost, latency, or capability.
Human-in-the-Loop (Coming Soon)
Approval gates for actions with external side effects — email, payments, API writes. Every escalation will be auditable and resumable. Currently in design.
Tool Plugin System (Planned)
A typed plugin interface to extend agent capabilities, with idempotent execution and structured observability. Internal abstractions exist; a public plugin API is on the roadmap.
Deterministic Workflows (Planned)
Stage-gated execution graphs for production workflows. Crash-safe and auditable transitions are on the roadmap.
Observability Built-In
Every tool execution, LLM call, and workflow transition emits structured logs with latency, token counts, and cost estimates out of the box.
Prompt Versioning
Version-controlled prompt templates per organization with rollback support. Changes are tracked and reviewed like code changes.
Usage Tracking
Rate limiting and quota enforcement built into the middleware layer. Usage metering and billing logic are planned for a future release.
Run it against your own agents.
Run the memory layer on your own infrastructure.
Start free on the hosted instance, or self-host the AGPLv3 core in your own VPC. No credit card, no sales call.