Episode and memory separation
Stores raw events as append-only episodes and compiles them into typed semantic memories, preserving both the event timeline and durable facts.
Statewave
Freemium
Statewave is an open-source, self-hosted memory runtime for AI agents. It stores durable episodes, compiles structured memories with provenance, and returns ranked context bundles for applications that need continuity across sessions.
Statewaveとは？
Statewave is an open-source memory runtime for AI agents. It provides a persistence layer between an LLM application and the users, projects, decisions, and events it encounters over time. Applications send raw events as immutable episodes scoped to subjects such as users, accounts, agents, repositories, or projects; Statewave compiles those episodes into typed memories and assembles relevant context for later requests.
The runtime separates episodic records from semantic memories. Episodes preserve what happened, while compiled memories represent facts such as profiles, preferences, procedures, and prior-issue summaries. Memories can include confidence scores, validity windows, and links to source episodes. Retrieval ranks relevant memories and episodes using signals including recency, similarity, kind priority, and temporal validity, then returns a token-bounded context bundle for an LLM prompt.
Statewave is self-hosted on Postgres with pgvector and exposes REST, Python, and TypeScript interfaces. Its MCP server supports compatible clients, while optional connectors normalize events from development, communication, documentation, support, and workflow tools into the common episode format.
Statewaveでできること
Ranked, token-bounded context
Combines relevant memories and episodes into a ranked context bundle constrained by the token budget selected by the application.
Provenance and state-assembly receipts
Links compiled memories to source episodes and records each context call with an immutable, ULID-addressable receipt and integrity hash.
Governance controls
Supports sensitivity labels, operator review before promotion, declarative YAML policies, deny or redact behavior, and an audit-oriented log_only mode.
Self-hosted integration surface
Runs with Python and Postgres, exposes REST and SDK interfaces, and offers an MCP server plus modular source connectors.
利用シーン
“Session-aware support agents”
Retain customer conversations, resolution history, and repeat-issue context so support workflows can assemble relevant history during handoffs or new tickets.
“Coding and repository agents”
Ingest issues, pull requests, reviews, releases, and architecture notes so an agent can build project knowledge across coding sessions.
“Internal copilots”
Preserve user workflows, previous decisions, and organizational context for internal tools without replaying complete interaction histories into every prompt.
“Long-lived subject-based agents”
Give agents that operate over extended periods a persistent memory layer organized around subjects such as users, accounts, communities, contacts, or projects.
よくある質問
What does Statewave return to an application?
It returns ranked, token-bounded context bundles containing relevant episodic records and semantic memories. Applications can place the bundles in a system prompt, tool result, or retrieval-augmented workflow.
How is Statewave different from a vector database?
It uses pgvector underneath but adds typed memories, append-only episodes, deterministic ranking signals, idempotent compilation, temporal validity, conflict resolution, provenance, and context assembly.
Can Statewave be self-hosted?
Yes. It is designed for self-hosting with Postgres and pgvector. Docker Compose can run the stack locally, and deployments can use container platforms such as Fly.io or Railway. The site states that there is no managed Statewave cloud.
Which models and frameworks can it work with?
Statewave is framework-neutral. Compilation uses LiteLLM, which supports numerous hosted and local providers, while retrieval and context assembly are provider-agnostic. REST, Python, and TypeScript interfaces support custom agents and frameworks.
Is Statewave open source for commercial use?
The project is released under Apache License 2.0, which permits use, modification, distribution, self-hosting, and commercial products. Optional enterprise support is offered as a services contract.
クイック情報
- Category
- AI infrastructure / developer tool
- Deployment
- Self-hosted Python service with Postgres and pgvector
- License
- Apache License 2.0
- Interfaces
- REST API, Python SDK, TypeScript SDK, and MCP server
- Core workflow
- Record episodes → compile memories → retrieve context bundles
- Source domain
- statewave.ai
Statewave のトラフィック分析
トラフィックデータは参考情報としてご利用ください。
- ドメイン評価
- 10
Statewaveの代替品
Cognee
www.cognee.ai
Cognee is an open-source memory platform for AI agents and applications. It connects data across sources, builds connected memory using vector, graph, and relational storage, and makes that context available through SDK, cloud, or self-hosted deployments.
Firetower
usefiretower.com
Firetower is an open-source, self-hosted control plane for running coding agents on servers you own. It lets developers start agent sessions from repositories, monitor them from desktop or mobile clients, review changes, and ship branches without keeping the initiating device online.
Flyte
flyte.org
An open-source runtime for durable AI workflows, model training, serving, and agents.
ByteRover
byterover.dev
ByteRoverはAIエージェントとチーム向けのコンテキストレイヤーです。プロジェクト知識をメモリーファイルとして保存し、セッション、エージェント、チームメンバー間の共有を管理できます。オンプレミスとプライベートVPCにも対応します。
Agents API
developers.openai.com
Agents API gives applications access to an OpenAI-managed Codex harness for building durable cloud agents. It manages sessions, orchestration, context compaction, and recovery while the application supplies tools and selects the execution environment.
iphone-use
github.com
Open-source, self-hosted iPhone control for AI agents on macOS.