Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
sage-wiki is a graph memory and knowledge base that AI agents and humans build and query together. Drop in documents; an LLM compiler turns them into an interlinked wiki with a knowledge graph. One Go binary scales it from a personal vault to a team hub to a company knowledge graph.
| Date | Stars |
|---|---|
| 2026-07-31 | 568 |
| 2026-08-02 | 569 |
| 2026-08-06 | 569 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
60.0
growth rate 0.00%/day
**English** | [中文](docs/translations/README_zh.md) | [日本語](docs/translations/README_ja.md) | [한국어](docs/translations/README_ko.md) | [Tiếng Việt](docs/translations/README_vi.md) | [Français](docs/translations/README_fr.md) | [Русский](docs/translations/README_ru.md) # sage-wiki **sage-wiki** is a graph memory and knowledge base that AI agents and humans build and query together. Drop in documents; an LLM compiler turns them into an interlinked wiki with a knowledge graph — agents query it through MCP, humans browse it as plain markdown. Enable the opt-in graph passes and it becomes an *evidenced* graph: typed entities, provenance-bearing relations, resolved aliases, and per-fact citations on answers. One Go binary scales it from a personal vault to a team hub to a company knowledge graph. **→ Get started: [Install](#install) · [Quickstart](#quickstart)** Grown from [Andrej Karpathy's idea](https://x.com/karpathy/status/2039805659525644595) of an LLM-compiled personal knowledge base, built with the [Sage Framework](https://github.com/xoai/sage). Some lessons learned along the way [here](https://x.com/xoai/status/2040936964799795503). - **Graph memory with citations.** Ask relational questions through `wiki_graph_query` — answers are grounded only in serialized graph edges; with the evidenced graph enabled, each citation carries its source document and confidence. - **Built for agents and humans.** 19 MCP tools plus generated skill files teach agents when to search, capture, and compile; humans get Obsidian-native markdown, a TUI, and a web UI over the same data. - **Trust and provenance.** Query outputs quarantine until verified; every evidenced relation records which document asserted it. - **Your sources in, a wiki out.** The compile pipeline reads papers, notes, code, and email; summarizes; extracts concepts; and writes interconnected articles — the ingestion layer for everything above. Every new source enriches existing articles; the wiki compounds as it grows. - **Ask your wiki questions.** Hybrid chunk-level search with LLM query expansion, re-ranking, and graph-aware context assembly returns cited answers. - **Scales to 100K+ documents.** Tiered compilation indexes everything fast and spends LLM budget only where it matters. https://github.com/user-attachments/assets/c35ee202-e9df-4ccd-b520-8f057163ff26 _Dots on the outer boundary represent summaries of all documents in the knowledge base, while dots in the inner circle represent concepts extracted from the knowledge base, with links showing how those concepts connect to one another._ ## From personal vault to company knowledge graph - **Personal** — overlay an existing Obsidian vault (`init --vault`), run on [local models](docs/guides/local-models.md) for zero cost, and opt into the graph passes (`ontology.triples` + `ontology.resolve`) when you want the evidenced graph. - **Team** — share one wiki via git or a [self-hosted server](docs/guides/self-hosted-server.md), review entity-resolution proposals and [output trust](docs/guides/output-trust.md) together, and federate multiple wikis with the hub. See [Team Setup](docs/guides/team-setup.md). - **Company** — move storage to [PostgreSQL/pgvector](docs/guides/storage-backends.md), turn on [metrics](docs/guides/metrics.md), front the server with auth, and scale ingestion with [tiered compilation](docs/guides/large-vault-performance.md). ## Knowledge graph & graph memory  Vector search retrieves passages that *look like* the query. A graph also records **how things relate**, so a question needing two or three hops is answered by traversal instead of hoping one chunk happens to contain the whole chain. sage-wiki builds that graph as a compile output — not a second database you have to keep in sync. - **Entities and typed relations.** Each compile extracts entities (concepts, sources, artifacts) and links them with typed relations. The relation voca
Excerpt of 30,953 characters
Read on GitHub
800
16
10
cyxer000
6
4
2
1
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:478f4cdbecf0434c, desc:knowledge base, desc:knowledge graph