---
README.md | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/README.md b/README.md
index 371bf0cd3..17b3328bd 100644
--- a/README.md
+++ b/README.md
@@ -1,5 +1,10 @@
+The Knowledge Layer for AI Agents
+
+ Prepare your unstructured data for AI Agents with high-performance ingestion and parsing.
+
+
---
From f1d369ed9ba2aed4a03e9bdf06013f031c8f76dd Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:26:05 +0800
Subject: [PATCH 14/52] docs: simplify tagline to a single centered heading
Co-authored-by: Cursor
---
README.md | 5 +----
1 file changed, 1 insertion(+), 4 deletions(-)
diff --git a/README.md b/README.md
index 17b3328bd..628bd9ce6 100644
--- a/README.md
+++ b/README.md
@@ -1,9 +1,6 @@
-The Knowledge Layer for AI Agents
-
- Prepare your unstructured data for AI Agents with high-performance ingestion and parsing.
-
+Prepare unstructured data for AI Agents
---
From f5e0fec274200be9bd7c3a1375df95e485359294 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:26:37 +0800
Subject: [PATCH 15/52] docs: remove horizontal line between tagline and badges
Co-authored-by: Cursor
---
README.md | 2 --
1 file changed, 2 deletions(-)
diff --git a/README.md b/README.md
index 628bd9ce6..66b9fb9b9 100644
--- a/README.md
+++ b/README.md
@@ -2,8 +2,6 @@
Prepare unstructured data for AI Agents
----
-
From f33e8210d9cb16ff9aa71514a97b2d90e9536053 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:28:54 +0800
Subject: [PATCH 16/52] docs: update news to reflect full open source and
self-hosted option
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 66b9fb9b9..9d6a5567c 100644
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@
## π’ News
-- **May 7, 2026**: π **Knowhere is now Open Source!** We've opened up our backend repository for document ingestion, parsing, and MCP-oriented knowledge access. Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
+- **May 7, 2026**: π **Knowhere is now Open Source!** We've opened up our entire stack for document ingestion, parsing, and MCP-oriented knowledge access. For those who prefer to run it themselves, you can now deploy the full experience using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the MCP knowledge access layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
Knowhere API is the backend repository for document ingestion, parsing,
From 85c57c96ea7c796b62f88ca13762913382b4a896 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:30:02 +0800
Subject: [PATCH 17/52] docs: use 'open-sourced' verb in news for better
clarity
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 9d6a5567c..062b14025 100644
--- a/README.md
+++ b/README.md
@@ -26,7 +26,7 @@
## π’ News
-- **May 7, 2026**: π **Knowhere is now Open Source!** We've opened up our entire stack for document ingestion, parsing, and MCP-oriented knowledge access. For those who prefer to run it themselves, you can now deploy the full experience using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
+- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and MCP-oriented knowledge access. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the MCP knowledge access layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
Knowhere API is the backend repository for document ingestion, parsing,
From 29878fda2813a79cb7d4f079dd585d3198cca4e2 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:30:35 +0800
Subject: [PATCH 18/52] docs: replace 'MCP-oriented knowledge' with 'agentic
RAG' for better technical accuracy
Co-authored-by: Cursor
---
README.md | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/README.md b/README.md
index 062b14025..585a86cf4 100644
--- a/README.md
+++ b/README.md
@@ -26,11 +26,11 @@
## π’ News
-- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and MCP-oriented knowledge access. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
-- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the MCP knowledge access layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
+- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
+- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the agentic RAG layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
Knowhere API is the backend repository for document ingestion, parsing,
-retrieval, and MCP-oriented knowledge access.
+retrieval, and agentic RAG.
## Features
From e11f74880afcdaae094031f056dc200fbe4d2357 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:32:58 +0800
Subject: [PATCH 19/52] docs: update high-level project description to
emphasize infrastructure role
Co-authored-by: Cursor
---
README.md | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/README.md b/README.md
index 585a86cf4..fe7dffeba 100644
--- a/README.md
+++ b/README.md
@@ -29,8 +29,7 @@
- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the agentic RAG layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
-Knowhere API is the backend repository for document ingestion, parsing,
-retrieval, and agentic RAG.
+Knowhere is the open-source infrastructure for unstructured data processing. It automates the complex pipeline of extracting, parsing, and transforming messy documents into structured, high-quality data optimized for AI Agents, agentic RAG, and traditional vector-based RAG workflows.
## Features
From 5b0d9869edcea5185b9c54086151a6aec79fefa8 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:34:05 +0800
Subject: [PATCH 20/52] docs: simplify project description to a single concise
sentence
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index fe7dffeba..e64d3071a 100644
--- a/README.md
+++ b/README.md
@@ -29,7 +29,7 @@
- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the agentic RAG layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
-Knowhere is the open-source infrastructure for unstructured data processing. It automates the complex pipeline of extracting, parsing, and transforming messy documents into structured, high-quality data optimized for AI Agents, agentic RAG, and traditional vector-based RAG workflows.
+Knowhere is the open-source data layer that makes unstructured documents ready for AI Agents and RAG.
## Features
From 6711651468a8cabe1f4ab790d2ca2e1c053d9632 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:34:48 +0800
Subject: [PATCH 21/52] docs: move project description below badges
Co-authored-by: Cursor
---
README.md | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/README.md b/README.md
index e64d3071a..c3e1a1a98 100644
--- a/README.md
+++ b/README.md
@@ -24,13 +24,13 @@
+Knowhere is the open-source data layer that makes unstructured documents ready for AI Agents and RAG.
+
## π’ News
- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the agentic RAG layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
-Knowhere is the open-source data layer that makes unstructured documents ready for AI Agents and RAG.
-
## Features
## Project Governance
From 4433d5903879ac99f25e595d6aec82267d63fe84 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:36:22 +0800
Subject: [PATCH 22/52] docs: move Project Governance section to bottom of
README
Co-authored-by: Cursor
---
README.md | 18 +++++++++---------
1 file changed, 9 insertions(+), 9 deletions(-)
diff --git a/README.md b/README.md
index c3e1a1a98..74b38d952 100644
--- a/README.md
+++ b/README.md
@@ -33,15 +33,6 @@ Knowhere is the open-source data layer that makes unstructured documents ready f
## Features
-## Project Governance
-
-- Licensed under Apache 2.0. See [LICENSE](LICENSE) and [NOTICE](NOTICE).
-- Contribution workflow and branch expectations live in
- [CONTRIBUTING.md](CONTRIBUTING.md).
-- Security reporting guidance lives in [SECURITY.md](SECURITY.md).
-- Community behavior expectations live in
- [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md).
-
## Repository Layout
```text
@@ -174,3 +165,12 @@ make check
- External dependency guide:
[docs/external-services.md](docs/external-services.md)
+
+## Project Governance
+
+- Licensed under Apache 2.0. See [LICENSE](LICENSE) and [NOTICE](NOTICE).
+- Contribution workflow and branch expectations live in
+ [CONTRIBUTING.md](CONTRIBUTING.md).
+- Security reporting guidance lives in [SECURITY.md](SECURITY.md).
+- Community behavior expectations live in
+ [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md).
From cb927ae5af46947230566b31bd749210b4e47aa3 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:40:06 +0800
Subject: [PATCH 23/52] docs: add navigation links for Website, Docs, and
Self-Host
Co-authored-by: Cursor
---
README.md | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/README.md b/README.md
index 74b38d952..ad09ea3bf 100644
--- a/README.md
+++ b/README.md
@@ -24,6 +24,12 @@
+
+ π Website |
+ π Docs |
+ π Self-Host
+
+
Knowhere is the open-source data layer that makes unstructured documents ready for AI Agents and RAG.
## π’ News
@@ -49,8 +55,6 @@ knowhere-api/
βββ build-images.yml
```
-## Architecture Overview
-
## Prerequisites
- Python 3.11+
@@ -59,8 +63,6 @@ knowhere-api/
- a local Chrome or Chromium driver if you plan to run document layout parsing
flows
-## Configuration
-
## Quick Start
1. Sync the workspace dependencies:
@@ -159,8 +161,6 @@ make check
- PostgreSQL: `localhost:5432`
- Redis: `localhost:6379`
-## Quick Example Request
-
## Additional Guides
- External dependency guide:
From e9b9ba6fe4b6408f07bdbd93368cf7db66c1a0f9 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:43:12 +0800
Subject: [PATCH 24/52] docs: restore detailed project description with focus
on infrastructure and RAG workflows
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index ad09ea3bf..e2bc69c80 100644
--- a/README.md
+++ b/README.md
@@ -30,7 +30,7 @@
π Self-Host
-Knowhere is the open-source data layer that makes unstructured documents ready for AI Agents and RAG.
+Knowhere is the open-source infrastructure for unstructured data processing. It automates the complex pipeline of extracting, parsing, and transforming messy documents into structured, high-quality data optimized for *AI Agents*, *Agentic RAG*, and *traditional vector-based RAG workflows*.
## π’ News
From 82a3d47f07229036d51e553f20f79141064db29f Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Wed, 6 May 2026 12:51:13 +0800
Subject: [PATCH 25/52] docs: add Knowhere Cloud CTA with free credits offer
Co-authored-by: Cursor
---
README.md | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/README.md b/README.md
index e2bc69c80..8a3de7355 100644
--- a/README.md
+++ b/README.md
@@ -32,6 +32,10 @@
Knowhere is the open-source infrastructure for unstructured data processing. It automates the complex pipeline of extracting, parsing, and transforming messy documents into structured, high-quality data optimized for *AI Agents*, *Agentic RAG*, and *traditional vector-based RAG workflows*.
+> [!NOTE]
+> **Get started in seconds with Knowhere Cloud.**
+> Avoid the complexity of self-deployment. Use our managed API at [knowhereto.ai](https://knowhereto.ai) and enjoy **$5 in free credits** upon registration.
+
## π’ News
- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
From 01ba66503a70288c5fcaecaeb848a3c8b2206941 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:06:41 +0800
Subject: [PATCH 26/52] docs: add 'How it Works' section and populate Features
Co-authored-by: Cursor
---
README.md | 53 +++++++++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 53 insertions(+)
diff --git a/README.md b/README.md
index 8a3de7355..4d79d1da3 100644
--- a/README.md
+++ b/README.md
@@ -41,8 +41,61 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
- **May 7, 2026**: π **Knowhere is now Open Source!** We have open-sourced our entire stack for document ingestion, parsing, and agentic RAG. You can now self-host the full platform using [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted). Check out our [Contribution Guide](CONTRIBUTING.md) to get involved!
- **Apr 30, 2026**: π¦ **Version [2026.04.30.1](https://github.com/Ontos-AI/knowhere/releases/tag/2026.04.30.1) has been released.** This update includes several stability improvements and initial support for the agentic RAG layer. See the [full changelog](https://github.com/Ontos-AI/knowhere/commits/2026.04.30.1) for details.
+## How it Works
+
+Knowhere has one simple goal: turn uploaded documents into a long-term memory store that agents can understand, navigate, and cite.
+
+The system can be understood as a three-stage pipeline:
+
+```text
+Document parsing -> Memory graph construction -> Agentic retrieval -> Cited results
+```
+
+### 1. Document Parsing
+
+Knowhere first profiles each uploaded file and routes it to the right parser: regular PDF, scanned drawing, Word, PowerPoint, spreadsheet, image, Markdown, or plain text. The parser converts the original file into structured content units:
+
+- `text`: paragraphs, headings, and body content
+- `table`: table content and table asset references
+- `image`: image content, OCR, or visual summaries
+- `path`: the hierarchical location of each chunk in the source document
+- `metadata`: summaries, keywords, media references, and contextual links between knowledge chunks and documents of various formats
+
+The parser is designed to preserve document structure, not just extract raw text. Heading hierarchy, section paths, images, and tables are converted into chunks that can be searched, navigated, and cited.
+
+### 2. Memory Graph Construction
+
+After parsing, Knowhere publishes the chunks into a canonical retrieval state:
+
+- `Document` represents a user document.
+- `DocumentSection` stores the document's section hierarchy.
+- `DocumentChunk` stores the final content units returned to users or agents.
+- `GraphNode` stores document-level memory, including summaries, keywords, chunk counts, media types, and navigation sections.
+- `GraphEdge` stores relationships between documents.
+
+The graph is intentionally lightweight. Its purpose is not to create an overly complex ontology, but to help the system understand which documents may be related, what each document is about, and why neighboring documents might matter. The current graph centers on document-level nodes and builds connections from keyword overlap, summaries, and navigation structure, making it useful for fast routing and expansion during retrieval.
+
+### 3. Agentic Retrieval
+
+At query time, Knowhere combines bottom-layer retrieval with agent-guided navigation:
+
+- **Bottom discovery**: retrieve candidate chunks from path, content, and term channels, then fuse them with RRF.
+- **Document selection**: the agent reads the memory graph overview and selects potentially relevant documents.
+- **Path selection**: the agent reviews compact chunk previews and selects the most relevant section or chunk paths.
+- **Hierarchical navigation**: for large documents, the agent takes a progressive strategy; it first inspects top-level and second-level sections, then expands selected sections to leaf chunks.
+- **Result merging**: agent-selected paths and bottom-layer candidates are merged, deduplicated, ranked, and returned with source citations.
+
+This gives Knowhere the stability of traditional retrieval and the structural awareness of agent navigation. The final output is not just a text snippet. It is cited evidence: the source document, section, chunk, and, when needed, linked image or table assets.
+
+> **TL;DR**: Knowhere parses documents into structured memory units, organizes them with a lightweight graph, and lets agents navigate through graph context and section paths to find evidence that can be reliably cited.
+
## Features
+- **Multi-modal Parsing**: High-fidelity extraction from PDF, Office, and images, preserving headings, tables, and hierarchical paths.
+- **Lightweight Memory Graph**: Context-aware organization that links documents and chunks for better relationship understanding.
+- **Agentic RAG**: A hybrid retrieval engine combining traditional search (RRF) with autonomous agent navigation.
+- **Evidence-based Citations**: Every result is backed by traceable source paths, ensuring reliability for AI Agent decision-making.
+
## Repository Layout
```text
From c7f4b14eafbaa6c32b89d067357d66fc671b66ab Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:08:20 +0800
Subject: [PATCH 27/52] docs: polish 'How it Works' section for better
readability and tone
Co-authored-by: Cursor
---
README.md | 52 ++++++++++++++++------------------------------------
1 file changed, 16 insertions(+), 36 deletions(-)
diff --git a/README.md b/README.md
index 4d79d1da3..45af781c6 100644
--- a/README.md
+++ b/README.md
@@ -43,51 +43,31 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
## How it Works
-Knowhere has one simple goal: turn uploaded documents into a long-term memory store that agents can understand, navigate, and cite.
-
-The system can be understood as a three-stage pipeline:
+Knowhere turns raw documents into a structured memory store that AI agents can navigate and cite. The process follows a three-stage pipeline:
```text
-Document parsing -> Memory graph construction -> Agentic retrieval -> Cited results
+Parsing -> Graph Construction -> Agentic Retrieval
```
### 1. Document Parsing
+Knowhere routes files to specialized parsers for PDFs, Office docs, images, and more. We don't just extract text; we preserve the document's hierarchy:
+- **Hierarchical Paths**: Every chunk knows its exact location (e.g., `Section 2.1 > Table 4`).
+- **Multi-modal Units**: Tables and images are treated as distinct assets with their own metadata.
+- **Structural Awareness**: Heading levels and section boundaries are maintained to keep context intact.
-Knowhere first profiles each uploaded file and routes it to the right parser: regular PDF, scanned drawing, Word, PowerPoint, spreadsheet, image, Markdown, or plain text. The parser converts the original file into structured content units:
-
-- `text`: paragraphs, headings, and body content
-- `table`: table content and table asset references
-- `image`: image content, OCR, or visual summaries
-- `path`: the hierarchical location of each chunk in the source document
-- `metadata`: summaries, keywords, media references, and contextual links between knowledge chunks and documents of various formats
-
-The parser is designed to preserve document structure, not just extract raw text. Heading hierarchy, section paths, images, and tables are converted into chunks that can be searched, navigated, and cited.
-
-### 2. Memory Graph Construction
-
-After parsing, Knowhere publishes the chunks into a canonical retrieval state:
-
-- `Document` represents a user document.
-- `DocumentSection` stores the document's section hierarchy.
-- `DocumentChunk` stores the final content units returned to users or agents.
-- `GraphNode` stores document-level memory, including summaries, keywords, chunk counts, media types, and navigation sections.
-- `GraphEdge` stores relationships between documents.
-
-The graph is intentionally lightweight. Its purpose is not to create an overly complex ontology, but to help the system understand which documents may be related, what each document is about, and why neighboring documents might matter. The current graph centers on document-level nodes and builds connections from keyword overlap, summaries, and navigation structure, making it useful for fast routing and expansion during retrieval.
+### 2. Memory Graph
+Parsed content is organized into a lightweight graph. Itβs designed as a practical map for agents, not a complex ontology.
+- **Nodes**: Represent documents, sections, and chunks.
+- **Edges**: Map semantic relationships (keyword overlap, summaries) and structural links.
+This graph helps agents quickly understand what a document is about and which neighboring files might be relevant.
### 3. Agentic Retrieval
+Retrieval combines traditional search with autonomous navigation:
+- **Hybrid Discovery**: Fuses keyword and semantic search (RRF) for broad coverage.
+- **Agent Navigation**: An agent "walks" the graph, reviewing previews to drill down into specific sections.
+- **Cited Evidence**: Results are returned as traceable evidence, including the source document, section, and any linked assets.
-At query time, Knowhere combines bottom-layer retrieval with agent-guided navigation:
-
-- **Bottom discovery**: retrieve candidate chunks from path, content, and term channels, then fuse them with RRF.
-- **Document selection**: the agent reads the memory graph overview and selects potentially relevant documents.
-- **Path selection**: the agent reviews compact chunk previews and selects the most relevant section or chunk paths.
-- **Hierarchical navigation**: for large documents, the agent takes a progressive strategy; it first inspects top-level and second-level sections, then expands selected sections to leaf chunks.
-- **Result merging**: agent-selected paths and bottom-layer candidates are merged, deduplicated, ranked, and returned with source citations.
-
-This gives Knowhere the stability of traditional retrieval and the structural awareness of agent navigation. The final output is not just a text snippet. It is cited evidence: the source document, section, chunk, and, when needed, linked image or table assets.
-
-> **TL;DR**: Knowhere parses documents into structured memory units, organizes them with a lightweight graph, and lets agents navigate through graph context and section paths to find evidence that can be reliably cited.
+**TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
## Features
From f2acb88a6571263d867e259970c9270b934a5204 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:10:13 +0800
Subject: [PATCH 28/52] docs: highlight TL;DR with GitHub Alerts syntax
Co-authored-by: Cursor
---
README.md | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 45af781c6..75b85fe47 100644
--- a/README.md
+++ b/README.md
@@ -43,6 +43,8 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
## How it Works
+**TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
+
Knowhere turns raw documents into a structured memory store that AI agents can navigate and cite. The process follows a three-stage pipeline:
```text
@@ -67,7 +69,8 @@ Retrieval combines traditional search with autonomous navigation:
- **Agent Navigation**: An agent "walks" the graph, reviewing previews to drill down into specific sections.
- **Cited Evidence**: Results are returned as traceable evidence, including the source document, section, and any linked assets.
-**TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
+> [!TIP]
+> **TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
## Features
From d97f5468d1d00082e459d6549cc746ff502f9776 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:11:45 +0800
Subject: [PATCH 29/52] docs: remove duplicate TL;DR, keep only the highlighted
version
Co-authored-by: Cursor
---
README.md | 2 --
1 file changed, 2 deletions(-)
diff --git a/README.md b/README.md
index 75b85fe47..ab0eb0b4d 100644
--- a/README.md
+++ b/README.md
@@ -43,8 +43,6 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
## How it Works
-**TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
-
Knowhere turns raw documents into a structured memory store that AI agents can navigate and cite. The process follows a three-stage pipeline:
```text
From a9e03d88a16f7b61533df3cf4335a7a75285c306 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:12:31 +0800
Subject: [PATCH 30/52] docs: move TL;DR tip to top of 'How it Works', remove
duplicate
Co-authored-by: Cursor
---
README.md | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/README.md b/README.md
index ab0eb0b4d..a2145fe0c 100644
--- a/README.md
+++ b/README.md
@@ -43,6 +43,9 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
## How it Works
+> [!TIP]
+> **TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
+
Knowhere turns raw documents into a structured memory store that AI agents can navigate and cite. The process follows a three-stage pipeline:
```text
@@ -67,9 +70,6 @@ Retrieval combines traditional search with autonomous navigation:
- **Agent Navigation**: An agent "walks" the graph, reviewing previews to drill down into specific sections.
- **Cited Evidence**: Results are returned as traceable evidence, including the source document, section, and any linked assets.
-> [!TIP]
-> **TL;DR**: Knowhere parses documents into structured units, maps them in a graph, and lets agents navigate that context to find and cite reliable evidence.
-
## Features
- **Multi-modal Parsing**: High-fidelity extraction from PDF, Office, and images, preserving headings, tables, and hierarchical paths.
From 8ca000ed8e2b483c87aa8c034c8399f15fc23529 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:13:31 +0800
Subject: [PATCH 31/52] docs: replace text pipeline with Mermaid diagram
showing parallel retrieval paths
Co-authored-by: Cursor
---
README.md | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/README.md b/README.md
index a2145fe0c..d97656cb3 100644
--- a/README.md
+++ b/README.md
@@ -48,8 +48,13 @@ Knowhere is the open-source infrastructure for unstructured data processing. It
Knowhere turns raw documents into a structured memory store that AI agents can navigate and cite. The process follows a three-stage pipeline:
-```text
-Parsing -> Graph Construction -> Agentic Retrieval
+```mermaid
+flowchart LR
+ A[π Document Parsing] --> B[πΈοΈ Graph Construction]
+ B --> C[π€ Agentic Retrieval]
+ B --> D[π Vector-based RAG]
+ C --> E[β
Cited Results]
+ D --> E
```
### 1. Document Parsing
From 7cd3b0974ec70eb8b69a6962d963c28d97ac85f9 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:16:06 +0800
Subject: [PATCH 32/52] docs: add Vector-based RAG section alongside Agentic
Retrieval
Co-authored-by: Cursor
---
README.md | 16 +++++++++++-----
1 file changed, 11 insertions(+), 5 deletions(-)
diff --git a/README.md b/README.md
index d97656cb3..fd035ea83 100644
--- a/README.md
+++ b/README.md
@@ -69,11 +69,17 @@ Parsed content is organized into a lightweight graph. Itβs designed as a pract
- **Edges**: Map semantic relationships (keyword overlap, summaries) and structural links.
This graph helps agents quickly understand what a document is about and which neighboring files might be relevant.
-### 3. Agentic Retrieval
-Retrieval combines traditional search with autonomous navigation:
-- **Hybrid Discovery**: Fuses keyword and semantic search (RRF) for broad coverage.
-- **Agent Navigation**: An agent "walks" the graph, reviewing previews to drill down into specific sections.
-- **Cited Evidence**: Results are returned as traceable evidence, including the source document, section, and any linked assets.
+### 3a. Agentic Retrieval
+An agent navigates the memory graph to find evidence rather than relying on a single vector lookup:
+- **Hybrid Discovery**: Fuses keyword and semantic search (RRF) for broad first-pass coverage.
+- **Agent Navigation**: The agent "walks" the graph, reviewing section previews to drill down into the most relevant paths.
+- **Cited Evidence**: Results are returned as traceable evidence β source document, section, chunk, and any linked image or table assets.
+
+### 3b. Vector-based RAG
+For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere's parsed chunks plug directly into standard vector stacks:
+- **Dense Search**: Chunk embeddings stored in Qdrant, pgvector, or Milvus for fast ANN lookup.
+- **Sparse Search**: BM25 term index for keyword-sensitive queries.
+- **Multi-channel Fusion**: Dense and sparse results are fused with RRF before being returned, giving you the best of both signals.
## Features
From bdf3ed31a09eff33646f1e58898abfc5f1e64e6b Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:25:45 +0800
Subject: [PATCH 33/52] docs: remove Repository Layout section
Co-authored-by: Cursor
---
README.md | 16 ----------------
1 file changed, 16 deletions(-)
diff --git a/README.md b/README.md
index fd035ea83..05101b46e 100644
--- a/README.md
+++ b/README.md
@@ -88,22 +88,6 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
- **Agentic RAG**: A hybrid retrieval engine combining traditional search (RRF) with autonomous agent navigation.
- **Evidence-based Citations**: Every result is backed by traceable source paths, ensuring reliability for AI Agent decision-making.
-## Repository Layout
-
-```text
-knowhere-api/
-βββ apps/
-β βββ api/
-β βββ worker/
-βββ packages/
-β βββ shared-python/
-βββ deploy/
-β βββ docker/
-β βββ local-dev/
-βββ .github/workflows/
- βββ build-images.yml
-```
-
## Prerequisites
- Python 3.11+
From 7857f6158f5d0b1da6ec322e277b572e6b2083fd Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:31:02 +0800
Subject: [PATCH 34/52] docs: replace Project Governance with Communication and
Contribution sections
Co-authored-by: Cursor
---
README.md | 26 ++++++++++++++++++--------
1 file changed, 18 insertions(+), 8 deletions(-)
diff --git a/README.md b/README.md
index 05101b46e..c9bd24444 100644
--- a/README.md
+++ b/README.md
@@ -199,11 +199,21 @@ make check
- External dependency guide:
[docs/external-services.md](docs/external-services.md)
-## Project Governance
-
-- Licensed under Apache 2.0. See [LICENSE](LICENSE) and [NOTICE](NOTICE).
-- Contribution workflow and branch expectations live in
- [CONTRIBUTING.md](CONTRIBUTING.md).
-- Security reporting guidance lives in [SECURITY.md](SECURITY.md).
-- Community behavior expectations live in
- [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md).
+## Communication
+
+- [GitHub Discussions](https://github.com/Ontos-AI/knowhere/discussions) for questions, ideas, and general conversation.
+- [GitHub Issues](https://github.com/Ontos-AI/knowhere/issues) for bug reports and feature requests.
+
+## Contribution
+
+Any contributions to Knowhere are more than welcome!
+
+If you are new to the project, check out the [good first issues](https://github.com/Ontos-AI/knowhere/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22). They are well-defined, relatively simple, and a great way to get familiar with the codebase and the contribution workflow.
+
+For general guidelines on branching, commit conventions, and the review process, take a look at [CONTRIBUTING.md](CONTRIBUTING.md).
+
+Other useful references:
+
+- [SECURITY.md](SECURITY.md) β how to report vulnerabilities responsibly.
+- [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md) β community behavior expectations.
+- [LICENSE](LICENSE) and [NOTICE](NOTICE) β Apache 2.0.
From e3a69707355f6eb1bdcc743ad2f1f00da3d6965d Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:36:15 +0800
Subject: [PATCH 35/52] docs: add knowhere-dashboard link to nav and Quick
Start
Co-authored-by: Cursor
---
README.md | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/README.md b/README.md
index c9bd24444..f8594e109 100644
--- a/README.md
+++ b/README.md
@@ -27,7 +27,8 @@
π Website |
π Docs |
- π Self-Host
+ π Self-Host |
+ π₯οΈ Dashboard
Knowhere is the open-source infrastructure for unstructured data processing. It automates the complex pipeline of extracting, parsing, and transforming messy documents into structured, high-quality data optimized for *AI Agents*, *Agentic RAG*, and *traditional vector-based RAG workflows*.
@@ -160,6 +161,8 @@ cd apps/api && uv run main.py
cd apps/worker && uv run worker.py
```
+The API is now running at `http://localhost:5005`. If you want the full product experience with a UI, run the [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) alongside it β it connects to this API out of the box.
+
## Quality Checks
Run lint checks from the repository root:
From 747aca949117882bdc0c8142065dc1e37e01264f Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:37:37 +0800
Subject: [PATCH 36/52] docs: add Ecosystem table to CONTRIBUTING.md with all
related repos
Co-authored-by: Cursor
---
CONTRIBUTING.md | 12 +++++++++++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 5fa02ad4f..94242ca7f 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -1,6 +1,16 @@
# Contributing
-Thanks for contributing to Knowhere API.
+Thanks for contributing to Knowhere. The project is split across several repositories β make sure you're working in the right one.
+
+## Ecosystem
+
+| Repository | Description |
+|---|---|
+| [knowhere](https://github.com/Ontos-AI/knowhere) | **This repo.** Backend API and worker β document ingestion, parsing, graph construction, and retrieval. |
+| [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
+| [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
+| [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
+| [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
## Before You Start
From 53b45d6d5563391ed3c032390b7d13b35f83cd98 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:38:18 +0800
Subject: [PATCH 37/52] docs: add Ecosystem table to README
Co-authored-by: Cursor
---
README.md | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/README.md b/README.md
index f8594e109..88c660e1b 100644
--- a/README.md
+++ b/README.md
@@ -82,6 +82,16 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
- **Sparse Search**: BM25 term index for keyword-sensitive queries.
- **Multi-channel Fusion**: Dense and sparse results are fused with RRF before being returned, giving you the best of both signals.
+## Ecosystem
+
+| Repository | Description |
+|---|---|
+| [knowhere](https://github.com/Ontos-AI/knowhere) | **This repo.** Backend API and worker β document ingestion, parsing, graph construction, and retrieval. |
+| [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
+| [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
+| [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
+| [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
+
## Features
- **Multi-modal Parsing**: High-fidelity extraction from PDF, Office, and images, preserving headings, tables, and hierarchical paths.
From 496f1badd3f07e6c12faf6eeab8cb86f034b17a5 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:42:39 +0800
Subject: [PATCH 38/52] docs: add Supported Formats section with coming soon
list
Co-authored-by: Cursor
---
README.md | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/README.md b/README.md
index 88c660e1b..550a2ce2b 100644
--- a/README.md
+++ b/README.md
@@ -99,6 +99,14 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
- **Agentic RAG**: A hybrid retrieval engine combining traditional search (RRF) with autonomous agent navigation.
- **Evidence-based Citations**: Every result is backed by traceable source paths, ensuring reliability for AI Agent decision-making.
+## Supported Formats
+
+| Supported | Coming Soon |
+|---|---|
+| `.pdf` `.docx` `.pptx` `.xlsx` `.csv` | `.epub` `.html` `.xml` |
+| `.jpg` `.png` | `.mp4` `.mp3` |
+| `.md` `.txt` `.json` | `.skills.md` |
+
## Prerequisites
- Python 3.11+
From b2a8e28239b52a352fc3ba5899f54466549ddf66 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:43:56 +0800
Subject: [PATCH 39/52] docs: add contribution invite to Supported Formats
section
Co-authored-by: Cursor
---
README.md | 2 ++
1 file changed, 2 insertions(+)
diff --git a/README.md b/README.md
index 550a2ce2b..3f6f340cf 100644
--- a/README.md
+++ b/README.md
@@ -107,6 +107,8 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
| `.jpg` `.png` | `.mp4` `.mp3` |
| `.md` `.txt` `.json` | `.skills.md` |
+Want to see a new format supported? Contributions are very welcome β adding a parser is a great first contribution. Check out [CONTRIBUTING.md](CONTRIBUTING.md) to get started.
+
## Prerequisites
- Python 3.11+
From b51b780ab1d3c39f273f84a213fee34abd55c107 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:44:31 +0800
Subject: [PATCH 40/52] docs: remove em dash in Supported Formats invite
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 3f6f340cf..3052596b3 100644
--- a/README.md
+++ b/README.md
@@ -107,7 +107,7 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
| `.jpg` `.png` | `.mp4` `.mp3` |
| `.md` `.txt` `.json` | `.skills.md` |
-Want to see a new format supported? Contributions are very welcome β adding a parser is a great first contribution. Check out [CONTRIBUTING.md](CONTRIBUTING.md) to get started.
+Want to see a new format supported? Adding a parser is a great first contribution. Check out [CONTRIBUTING.md](CONTRIBUTING.md) to get started.
## Prerequisites
From 7cb3089d977297f7bbeb5112a2a85fdb285b425c Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:50:39 +0800
Subject: [PATCH 41/52] docs: add emoji to Supported Formats table headers
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 3052596b3..a3823db39 100644
--- a/README.md
+++ b/README.md
@@ -101,7 +101,7 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
## Supported Formats
-| Supported | Coming Soon |
+| β
Supported | β³ Coming Soon |
|---|---|
| `.pdf` `.docx` `.pptx` `.xlsx` `.csv` | `.epub` `.html` `.xml` |
| `.jpg` `.png` | `.mp4` `.mp3` |
From 23b8b85884d0e37e9a638a336cb4a15f2084dad3 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:55:04 +0800
Subject: [PATCH 42/52] docs: add emoji to Ecosystem tables in README and
CONTRIBUTING
Co-authored-by: Cursor
---
CONTRIBUTING.md | 8 ++++----
README.md | 8 ++++----
2 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 94242ca7f..dc81752ce 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -7,10 +7,10 @@ Thanks for contributing to Knowhere. The project is split across several reposit
| Repository | Description |
|---|---|
| [knowhere](https://github.com/Ontos-AI/knowhere) | **This repo.** Backend API and worker β document ingestion, parsing, graph construction, and retrieval. |
-| [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
-| [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
-| [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
-| [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
+| π₯οΈ [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
+| π³ [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
+| π [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
+| π¦ [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
## Before You Start
diff --git a/README.md b/README.md
index a3823db39..442effbe0 100644
--- a/README.md
+++ b/README.md
@@ -87,10 +87,10 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
| Repository | Description |
|---|---|
| [knowhere](https://github.com/Ontos-AI/knowhere) | **This repo.** Backend API and worker β document ingestion, parsing, graph construction, and retrieval. |
-| [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
-| [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
-| [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
-| [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
+| π₯οΈ [knowhere-dashboard](https://github.com/Ontos-AI/knowhere-dashboard) | The web UI. Connects to the API for the full product experience. |
+| π³ [knowhere-self-hosted](https://github.com/Ontos-AI/knowhere-self-hosted) | Docker Compose stack for self-hosted deployments. Packages the API, worker, and dashboard together. |
+| π [knowhere-python-sdk](https://github.com/Ontos-AI/knowhere-python-sdk) | Official Python SDK for the Knowhere Cloud API. |
+| π¦ [knowhere-node-sdk](https://github.com/Ontos-AI/knowhere-node-sdk) | Official Node.js SDK for the Knowhere Cloud API. |
## Features
From 58199c9141387b281ebb37e861dabb44c20a9d81 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:56:03 +0800
Subject: [PATCH 43/52] docs: switch Supported Formats to GitHub checklist
syntax
Co-authored-by: Cursor
---
README.md | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/README.md b/README.md
index 442effbe0..00309d4ec 100644
--- a/README.md
+++ b/README.md
@@ -101,11 +101,12 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
## Supported Formats
-| β
Supported | β³ Coming Soon |
-|---|---|
-| `.pdf` `.docx` `.pptx` `.xlsx` `.csv` | `.epub` `.html` `.xml` |
-| `.jpg` `.png` | `.mp4` `.mp3` |
-| `.md` `.txt` `.json` | `.skills.md` |
+- [x] `.pdf` `.docx` `.pptx` `.xlsx` `.csv`
+- [x] `.jpg` `.png`
+- [x] `.md` `.txt` `.json`
+- [ ] `.epub` `.html` `.xml`
+- [ ] `.mp4` `.mp3`
+- [ ] `.skills.md`
Want to see a new format supported? Adding a parser is a great first contribution. Check out [CONTRIBUTING.md](CONTRIBUTING.md) to get started.
From c61fe373818423ef179b21ad57514b948c2f2316 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:57:50 +0800
Subject: [PATCH 44/52] docs: split Supported Formats into labeled subsections
with emoji
Co-authored-by: Cursor
---
README.md | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/README.md b/README.md
index 00309d4ec..243965ca5 100644
--- a/README.md
+++ b/README.md
@@ -101,9 +101,14 @@ For teams that prefer a pure retrieval pipeline without agent overhead, Knowhere
## Supported Formats
+**β
Supported**
+
- [x] `.pdf` `.docx` `.pptx` `.xlsx` `.csv`
- [x] `.jpg` `.png`
- [x] `.md` `.txt` `.json`
+
+**β³ Coming Soon**
+
- [ ] `.epub` `.html` `.xml`
- [ ] `.mp4` `.mp3`
- [ ] `.skills.md`
From 7243ce29bce06ce15f4ef4b21ad70ee95309bfc1 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:59:23 +0800
Subject: [PATCH 45/52] docs: add Citation section to README and CITATION.cff
for GitHub native support
Co-authored-by: Cursor
---
CITATION.cff | 19 +++++++++++++++++++
README.md | 21 +++++++++++++++++++++
2 files changed, 40 insertions(+)
create mode 100644 CITATION.cff
diff --git a/CITATION.cff b/CITATION.cff
new file mode 100644
index 000000000..9e041ea19
--- /dev/null
+++ b/CITATION.cff
@@ -0,0 +1,19 @@
+cff-version: 1.2.0
+title: "Knowhere: The Open-Source Data Layer for AI Agents and RAG"
+message: "If you use this software in your research, please cite it using the metadata from this file."
+type: software
+authors:
+ - name: "Ontos AI"
+ website: "https://knowhereto.ai"
+repository-code: "https://github.com/Ontos-AI/knowhere"
+url: "https://knowhereto.ai"
+license: Apache-2.0
+version: "2026.04.30.1"
+date-released: "2026-04-30"
+keywords:
+ - RAG
+ - agentic RAG
+ - document parsing
+ - unstructured data
+ - AI agents
+ - knowledge graph
diff --git a/README.md b/README.md
index 243965ca5..945f3b6c0 100644
--- a/README.md
+++ b/README.md
@@ -228,6 +228,27 @@ make check
- External dependency guide:
[docs/external-services.md](docs/external-services.md)
+## Citation
+
+If you use Knowhere in your research, please cite it as:
+
+
+BibTeX
+
+```bibtex
+@software{knowhere2026,
+ author = {Ontos AI},
+ title = {Knowhere: The Open-Source Data Layer for AI Agents and RAG},
+ year = {2026},
+ publisher = {GitHub},
+ url = {https://github.com/Ontos-AI/knowhere},
+ version = {2026.04.30.1},
+ license = {Apache-2.0}
+}
+```
+
+
+
## Communication
- [GitHub Discussions](https://github.com/Ontos-AI/knowhere/discussions) for questions, ideas, and general conversation.
From d0b9c9e91162ec04b6ac8eb5d14bab5f0fa6cf82 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 12:59:56 +0800
Subject: [PATCH 46/52] docs: update citation title to match project tagline
Co-authored-by: Cursor
---
CITATION.cff | 2 +-
README.md | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/CITATION.cff b/CITATION.cff
index 9e041ea19..b3385f123 100644
--- a/CITATION.cff
+++ b/CITATION.cff
@@ -1,5 +1,5 @@
cff-version: 1.2.0
-title: "Knowhere: The Open-Source Data Layer for AI Agents and RAG"
+title: "Knowhere: Prepare Unstructured Data for AI Agents"
message: "If you use this software in your research, please cite it using the metadata from this file."
type: software
authors:
diff --git a/README.md b/README.md
index 945f3b6c0..4babf94e0 100644
--- a/README.md
+++ b/README.md
@@ -238,7 +238,7 @@ If you use Knowhere in your research, please cite it as:
```bibtex
@software{knowhere2026,
author = {Ontos AI},
- title = {Knowhere: The Open-Source Data Layer for AI Agents and RAG},
+ title = {Knowhere: Prepare Unstructured Data for AI Agents},
year = {2026},
publisher = {GitHub},
url = {https://github.com/Ontos-AI/knowhere},
From 6f497a8e1486146aa892dad82dbd53290c9c88d8 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:01:13 +0800
Subject: [PATCH 47/52] docs: remove collapsible wrapper from citation BibTeX
block
Co-authored-by: Cursor
---
README.md | 5 -----
1 file changed, 5 deletions(-)
diff --git a/README.md b/README.md
index 4babf94e0..4d7bef86d 100644
--- a/README.md
+++ b/README.md
@@ -232,9 +232,6 @@ make check
If you use Knowhere in your research, please cite it as:
-
-BibTeX
-
```bibtex
@software{knowhere2026,
author = {Ontos AI},
@@ -247,8 +244,6 @@ If you use Knowhere in your research, please cite it as:
}
```
-
-
## Communication
- [GitHub Discussions](https://github.com/Ontos-AI/knowhere/discussions) for questions, ideas, and general conversation.
From 35970bbeee5ebd52a0c51e405ae38ae37d0de496 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:02:29 +0800
Subject: [PATCH 48/52] docs: add We're Hiring section at the end of README
Co-authored-by: Cursor
---
README.md | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/README.md b/README.md
index 4d7bef86d..7efbd5741 100644
--- a/README.md
+++ b/README.md
@@ -262,3 +262,9 @@ Other useful references:
- [SECURITY.md](SECURITY.md) β how to report vulnerabilities responsibly.
- [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md) β community behavior expectations.
- [LICENSE](LICENSE) and [NOTICE](NOTICE) β Apache 2.0.
+
+---
+
+## We're Hiring
+
+We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out at [team@knowhereto.ai](mailto:team@knowhereto.ai).
From 41461b27fbe520b318bba48dedec4177f01e7061 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:03:51 +0800
Subject: [PATCH 49/52] docs: obfuscate hiring email with base64 encoding
Co-authored-by: Cursor
---
README.md | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 7efbd5741..1bfd178cc 100644
--- a/README.md
+++ b/README.md
@@ -267,4 +267,10 @@ Other useful references:
## We're Hiring
-We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out at [team@knowhereto.ai](mailto:team@knowhereto.ai).
+We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out β decode the address below and drop us a line:
+
+```
+dGVhbUBrbm93aGVyZXRvLmFp
+```
+
+`echo 'dGVhbUBrbm93aGVyZXRvLmFp' | base64 --decode`
From 34080b6b45e0cb9a7f4bed7793ea259e5bb52fe4 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:04:44 +0800
Subject: [PATCH 50/52] docs: remove divider and update hiring section heading
style
Co-authored-by: Cursor
---
README.md | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/README.md b/README.md
index 1bfd178cc..bcdaf23e4 100644
--- a/README.md
+++ b/README.md
@@ -263,9 +263,7 @@ Other useful references:
- [CODE_OF_CONDUCT.md](CODE_OF_CONDUCT.md) β community behavior expectations.
- [LICENSE](LICENSE) and [NOTICE](NOTICE) β Apache 2.0.
----
-
-## We're Hiring
+## π We're Hiring!
We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out β decode the address below and drop us a line:
From 8be11bc61057a93714de0736040c0cab3e57a5b4 Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:05:39 +0800
Subject: [PATCH 51/52] docs: simplify hiring email obfuscation to a single
bash command
Co-authored-by: Cursor
---
README.md | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/README.md b/README.md
index bcdaf23e4..37117baeb 100644
--- a/README.md
+++ b/README.md
@@ -267,8 +267,6 @@ Other useful references:
We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out β decode the address below and drop us a line:
-```
-dGVhbUBrbm93aGVyZXRvLmFp
-```
-
+```bash
`echo 'dGVhbUBrbm93aGVyZXRvLmFp' | base64 --decode`
+```
From f3f0773d688d8951cb39595ddee5cc66489dd2bf Mon Sep 17 00:00:00 2001
From: Steve Kwok
Date: Thu, 7 May 2026 13:06:15 +0800
Subject: [PATCH 52/52] docs: remove backtick wrapping from bash decode command
Co-authored-by: Cursor
---
README.md | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/README.md b/README.md
index 37117baeb..d744750d8 100644
--- a/README.md
+++ b/README.md
@@ -268,5 +268,5 @@ Other useful references:
We're building the knowledge layer for the Agent era. If that sounds like work you want to do, reach out β decode the address below and drop us a line:
```bash
-`echo 'dGVhbUBrbm93aGVyZXRvLmFp' | base64 --decode`
+echo 'dGVhbUBrbm93aGVyZXRvLmFp' | base64 --decode
```