MCPcopy Index your code
hub / github.com/K-Dense-AI/k-dense-byok

github.com/K-Dense-AI/k-dense-byok @v0.7.0

Chat with this repo
repository ↗ · DeepWiki ↗ · release v0.7.0 ↗ · + Follow
950 symbols 2,447 edges 177 files 50 documented · 5%
What it actually does AI analysis from the code graph — generated when you open this
loading…
README

K-Dense BYOK

License: MIT Version Skills Workflows Databases Tests X LinkedIn YouTube

Your own AI research assistant, running on your computer, powered by your API keys.

K-Dense BYOK — Kady running an end-to-end single-cell RNA-seq analysis, with the project file browser, a rendered analysis report, and the live chat panel

K-Dense BYOK (Bring Your Own Keys) is a free, open-source app that gives you Kady — an AI research assistant for scientists in any field. Describe a task in plain language — analyze a dataset, review a manuscript, search the literature, build a figure — and Kady does the work: it reads and writes files, runs real code, searches the web, and hands you the results.

No coding experience is required. The app runs locally on your machine, your data stays with you, and you pay only for the AI model usage on your own account.

Beta: K-Dense BYOK is currently in beta. Many features and improvements are on the way. Star us on GitHub to stay in the loop, and follow K-Dense on X, LinkedIn, and YouTube for release notes and tutorials.

What can it do?

  • Take on real research tasks — data analysis, literature review, manuscript checking, figure generation — in a full working environment, with progress shown live in the chat.
  • Delegate to a built-in team of 21 scientific specialists, like a citation-checker, a statistical-reviewer, and a peer-reviewer — running one at a time, in parallel, or chained. Learn more.
  • Search the web and read sources natively — pages, PDFs, GitHub repositories, even YouTube videos — with no extra API key required.
  • Ask before it assumes. When a task is ambiguous, Kady shows a quick question form in the chat instead of guessing.
  • Use 140+ pre-installed scientific skills covering genomics, proteomics, drug discovery, materials science, and more — activated automatically per task.
  • Launch 326 ready-made workflow templates across 22 disciplines: pick one, fill in the blanks, go.
  • Query 229 scientific and financial databases in 18 categories, from PubMed-scale biomedical resources to market and climate data.
  • Use any tool-capable AI model — OpenAI, Anthropic, Google, xAI, Qwen, and more via one OpenRouter account, or free local models via Ollama. Switch per chat.
  • Fuse multiple frontier models with OpenRouter Fusion — pick a Fusion preset (e.g. Opus 4.8 + GPT-5.5 + Gemini 3.1 Pro) and a panel of models deliberates while a judge synthesizes one answer, with combined pricing and DRACO benchmark scores shown right in the model picker.
  • Preview 60+ scientific file formats right in the app — interactive 3D protein structures, 2D chemical structures from SMILES, mass spectra, single-cell/HDF5/Parquet arrays, phylogenetic trees, sequence alignments, DICOM/NIfTI/microscopy imaging, and more, alongside the everyday CSVs, PDFs, notebooks, and genomics tables.
  • Stay organized with projects — each with its own files, chat history, up to 10 parallel chat tabs, a LaTeX editor with inline compile diagnostics, and cost tracking with optional spend caps.
  • Pick up where you left off — reopen any past chat from the session history menu, full transcript included, right alongside your open tabs.
  • Offload heavy or GPU work to the cloud with Modal — add your Modal keys and Kady can run a command or script on an on-demand sandbox, from a cheap CPU box up to H100 GPUs. Files sync back to your project and the compute cost is tracked right alongside model spend. Pick the instance per chat from the compute selector.
  • Extend it with MCP servers — connect GitHub, reference managers, databases, and hundreds of other external tools.
  • Manage everything from Settings — a built-in capability hub with panels to browse and toggle skills, create or disable specialists, and add or disconnect MCP connectors per project, plus live API key management (OpenRouter, Modal, and more) — no config files to edit, and disabling anything is non-destructive.

Get started in 5 minutes

You need a computer running macOS or Linux (Windows works via WSL) and an OpenRouter API key (one account, all major AI models — or use free local models instead).

Open a terminal and run:

git clone https://github.com/K-Dense-AI/k-dense-byok.git
cd k-dense-byok
cp .env.example .env    # then paste your OpenRouter key into .env
./start.sh

The first start installs everything automatically (it takes a few minutes); then your browser opens to http://localhost:3000. Press Ctrl+C in the terminal to stop. You can also add or change API keys anytime from the in-app Settings — no restart needed.

That's it. Create a project, drop in your data, and ask Kady for what you want — for example: "Run a differential expression analysis on counts.csv comparing treated vs control, and plot a volcano plot."

➡️ Step-by-step details, optional API keys, and troubleshooting: Installation guide ➡️ Your first session and everyday features: Basic usage

Documentation

All guides live in the docs/ folder:

Guide What it covers
Installation Full setup walkthrough, optional API keys, updating, troubleshooting
Basic usage First session, chat tabs, files, workflows, databases, costs, tips
File previews Every scientific format Kady can render — structures, spectra, imaging, arrays, and more
Sub-agents Kady's team of 21 scientific specialists and how to customize them
Connecting external tools (MCP) Give Kady extra abilities like GitHub, reference managers, and databases
Local models with Ollama Run everything on free local models, no API keys required
Model selection How Kady builds the OpenRouter model list
OpenRouter Fusion Multi-model deliberation presets (this fork) — what they are and how the integration works
Architecture How the two local services fit together
Contributing workflows Add new workflow templates to the library
Known limitations Rough edges to be aware of in the current beta

What's coming

Literature search (Paperclip), document conversion, browser automation, and the provenance/"Copy as Methods" export are being re-added in upcoming releases, along with MCP tools for sub-agents and a UI experience further tailored to scientific workflows. In the meantime, many of these are available today by connecting an MCP server.

Want more?

K-Dense BYOK is great for getting started, but if you want end-to-end research workflows with managed infrastructure, team collaboration, and no setup required, check out K-Dense Web — our full platform built for professional and academic research teams.

Issues, bugs, or feature requests

If you run into a problem or have an idea for something new, please open a GitHub issue. We read every one.

About K-Dense

K-Dense BYOK is open source because K-Dense believes in giving back to the community that makes this kind of work possible.

Star history

Star History Chart

Extension points exported contracts — how you extend this code

SubagentRunDetails (Interface)
Shape of the pi-subagents tool result details we consume (subset).
server/src/agent/subagent-bridge.ts
HistoryItem (Interface)
One transcript entry from GET /sessions/:id/history.
web/src/lib/use-agent.ts
AsyncCompletePayload (Interface)
Subset of the async completion payload (the runner's result-file JSON).
server/src/agent/subagent-bridge.ts
TreeNodesProps (Interface)
(no doc)
web/src/components/sandbox-panel.tsx
ProjectMeta (Interface)
(no doc)
server/src/projects.ts
FileTreePanelProps (Interface)
(no doc)
web/src/components/sandbox-panel.tsx
ProjectPaths (Interface)
(no doc)
server/src/projects.ts
ProjectFormState (Interface)
(no doc)
web/src/components/project-switcher.tsx

Core symbols most depended-on inside this repo

cn
called by 239
web/src/lib/utils.ts
apiFetch
called by 47
web/src/lib/projects.ts
activePaths
called by 32
server/src/projects.ts
runSciHelper
called by 32
server/src/api/sci-helpers.ts
ensureProjectExists
called by 30
server/src/projects.ts
resolvePaths
called by 27
server/src/projects.ts
fileCategory
called by 27
web/src/lib/use-sandbox.ts
formatUsd
called by 21
web/src/lib/utils.ts

Shape

Function 777
Interface 155
Method 11
Class 7

Languages

TypeScript94%
Python6%

Modules by API surface

web/src/components/ai-elements/prompt-input.tsx49 symbols
web/src/components/file-preview-panel.tsx46 symbols
web/src/components/interview-form.tsx24 symbols
server/src/projects.ts24 symbols
server/src/agent/mcp.ts23 symbols
server/src/agent/agent-files.ts22 symbols
web/src/components/ai-elements/speech-input.tsx20 symbols
web/src/components/viewers/arraydata-viewer.tsx16 symbols
web/src/components/ai-elements/message.tsx16 symbols
server/src/cost/ledger.ts16 symbols
web/src/components/ui/dropdown-menu.tsx15 symbols
server/src/helpers/imaging_helper.py15 symbols

For agents

$ claude mcp add k-dense-byok \
  -- python -m otcore.mcp_server <graph>

⬇ download graph artifact

Ask about this repo answers extend the page