diff --git a/README.md b/README.md index 84b846d..f0a6793 100644 --- a/README.md +++ b/README.md @@ -18,9 +18,11 @@ cloud: everything is derived from your files, every claim can quote the sentence it came from, and the whole graph rebuilds from the vault at any time. See [ARCHITECTURE.md](ARCHITECTURE.md) for the full design. -**Status: Phase 1 + vault intelligence** — vault CRUD, live entity extraction -with in-editor underlines, entity/relations panel, vault-wide graph -aggregation, and a deterministic enrichment layer: +**What's inside** — a Markdown vault with live entity extraction and +in-editor underlines, document ingestion (PDF/DOCX/HTML/TXT), hybrid +search, git-backed time travel, a cross-source chronology, an MCP server, +a Python library with publication-quality plots, and a deterministic +enrichment layer: - **Inferred connections** — a Datalog reasoner forward-chains rules over extracted facts: multi-hop chains ("Widget Inc ↔ Ada Lovelace, because diff --git a/backend/pyproject.toml b/backend/pyproject.toml index 1dc90cb..dfc635b 100644 --- a/backend/pyproject.toml +++ b/backend/pyproject.toml @@ -1,10 +1,10 @@ [project] name = "graphier" version = "0.1.0" -description = "An Obsidian-like knowledge workspace where the graph builds itself — no LLM, no cloud, every claim can quote its source sentence" +description = "Obsidian-like notes where the knowledge graph builds itself — no LLM, no cloud, every claim quotes its source sentence" license = { text = "MIT" } requires-python = ">=3.10" -keywords = ["knowledge-graph", "pkm", "note-taking", "local-first", "datalog", "obsidian"] +keywords = ["knowledge-graph", "pkm", "note-taking", "local-first", "datalog", "obsidian", "mcp-server", "entity-extraction", "provenance"] classifiers = [ "Development Status :: 4 - Beta", "Intended Audience :: End Users/Desktop", diff --git a/docs/launch-checklist.md b/docs/launch-checklist.md index 6dd8f0a..489c88e 100644 --- a/docs/launch-checklist.md +++ b/docs/launch-checklist.md @@ -5,10 +5,22 @@ CONTRIBUTING, ROADMAP, templates). What remains needs the repo owner. ## Repo settings (5 minutes) -- [ ] Topics: `knowledge-graph`, `pkm`, `note-taking`, `local-first`, - `obsidian-alternative`, `datalog`, `knowledge-management` -- [ ] Description: *"Obsidian-like notes where the knowledge graph builds - itself — no LLM, no cloud, every claim quotes its source sentence"* +On https://github.com/ioteverythin/Graphier, click the **gear** next to +"About" (top of the right sidebar) and paste — this cannot be set from a +PR, only the owner's UI/API: + +- [ ] Description: + + ``` + Obsidian-like notes where the knowledge graph builds itself — no LLM, no cloud, every claim quotes its source sentence. + ``` + +- [ ] Topics (comma/Enter separated): + + ``` + knowledge-graph, note-taking, pkm, obsidian, markdown, local-first, datalog, entity-extraction, mcp-server, self-hosted, fastapi, react, provenance, python + ``` + - [ ] Social preview image: upload `docs/graph-view.png` - [ ] Enable **Discussions** - [ ] Tag `v0.1.0` (the release workflow builds and attaches the wheel) @@ -44,8 +56,9 @@ CONTRIBUTING, ROADMAP, templates). What remains needs the repo owner. diff nodes/edges and render what appeared/vanished/changed. 7. **Obsidian vault import** — handle frontmatter and `#tags` when pointing GRAPHIER_VAULT at an existing Obsidian folder. -8. **MCP server** — expose search/entity/relations/Datalog/provenance to - MCP clients (roadmap headliner; design discussion first). +8. **Fuzzy / typo-tolerant search** — BM25 is exact-match; a cheap + edit-distance fallback for zero-hit queries would soften typos + (`search.py`, see docs/how-search-works.md for the constraints). ## Launch posts