What is this?
WeKnora is an open-source, self-hosted knowledge platform that turns raw documents into a queryable RAG knowledge base, an autonomous ReAct agent, and a self-maintaining Wiki with an interactive knowledge graph.
WeKnora is an open-source, self-hosted knowledge platform that turns raw documents into a queryable RAG knowledge base, an autonomous ReAct agent, and a self-maintaining Wiki with an interactive knowledge graph.Is this a library I import, or a full platform I deploy?It is a full self-hostable platform with a modular backend, web UI, and API, written in Go with a Docker Compose setup, not a code library you import into your own app.
It is a full self-hostable platform with a modular backend, web UI, and API, written in Go with a Docker Compose setup, not a code library you import into your own app.How do I deploy WeKnora?The project is designed for local or private-cloud deployment and ships with a Docker Compose setup, with full configuration docs on a dedicated VitePress documentation site.
The project is designed for local or private-cloud deployment and ships with a Docker Compose setup, with full configuration docs on a dedicated VitePress documentation site.Which LLM providers and backends can I plug in?WeKnora supports 20+ LLM providers including OpenAI, DeepSeek, Qwen, Zhipu, Hunyuan, Gemini, MiniMax, NVIDIA, LiteLLM, and Ollama, plus swappable vector databases (pgvector, Milvus, Qdrant, Weaviate, OpenSearch) and storage backends.
WeKnora supports 20+ LLM providers including OpenAI, DeepSeek, Qwen, Zhipu, Hunyuan, Gemini, MiniMax, NVIDIA, LiteLLM, and Ollama, plus swappable vector databases (pgvector, Milvus, Qdrant, Weaviate, OpenSearch) and storage backends.Are there any recent changes that affect how skill sandboxes run?As of v0.8.0, agent skill sandboxes run only on session-persistent Docker, E2B, or Cube backends; the older local host-process backend was removed and Docker execution is now opt-in rather than enabled by default.
As of v0.8.0, agent skill sandboxes run only on session-persistent Docker, E2B, or Cube backends; the older local host-process backend was removed and Docker execution is now opt-in rather than enabled by default.Does my data stay private if I self-host it?Yes — the architecture is designed for local or private cloud deployment so you keep full control over where documents and embeddings are stored.
Yes — the architecture is designed for local or private cloud deployment so you keep full control over where documents and embeddings are stored.What does Wiki Mode add beyond regular RAG answers?In Wiki Mode, agents distill raw documents into a self-maintaining, interlinked markdown knowledge base complete with an interactive knowledge graph, manual editing, revision history, and one-click rollback.
In Wiki Mode, agents distill raw documents into a self-maintaining, interlinked markdown knowledge base complete with an interactive knowledge graph, manual editing, revision history, and one-click rollback.Which chat apps can WeKnora answer questions through?It can serve Q&A directly inside IM channels like WeCom, Feishu, Slack, Telegram, DingTalk, Mattermost, and QQBot, in addition to its own web UI and embeddable website widget.
It can serve Q&A directly inside IM channels like WeCom, Feishu, Slack, Telegram, DingTalk, Mattermost, and QQBot, in addition to its own web UI and embeddable website widget.Can multiple teams share one WeKnora deployment?Yes — it has enterprise-ready multi-workspace RBAC with a 4-tier role matrix (Owner/Admin/Contributor/Viewer), per-resource ownership, and per-workspace audit logs.
Yes — it has enterprise-ready multi-workspace RBAC with a 4-tier role matrix (Owner/Admin/Contributor/Viewer), per-resource ownership, and per-workspace audit logs.How can I inspect what the agent is actually doing?WeKnora integrates with Langfuse for full observability into agent reasoning, token usage, and pipeline tracing, plus a runtime task-queue dashboard for worker-pool governance.
WeKnora integrates with Langfuse for full observability into agent reasoning, token usage, and pipeline tracing, plus a runtime task-queue dashboard for worker-pool governance.What are similar projects?goclaw, open-webui, LightRAG, vault-ai, casdoor, m_flow, ruflo, LLMSearchRecommender
goclaw, open-webui, LightRAG, vault-ai, casdoor, m_flow, ruflo, LLMSearchRecommenderWhich topics is it in?reranking, notion, question-answering, multi-tenant, wiki, agentic, vector-search, rbac, slack, knowledge-graph, vector-database, semantic-search
reranking, notion, question-answering, multi-tenant, wiki, agentic, vector-search, rbac, slack, knowledge-graph, vector-database, semantic-search