K-Dense BYOK: Open-Source AI Co-Scientist Runs Locally With a Tamper-Proof Lab Notebook
RexDouglass · x · 2026-10-03
The K-Dense team published an arXiv paper describing K-Dense BYOK, a free, open-source AI research assistant that runs entirely on the researcher's own computer, with the user supplying any model (hosted or local) while the app provides the execution environment, scientific scaffolding, and a complete record. Everything lives in ordinary folders, so data, code, results, and logs remain readable years later without the app.
Three things distinguish it from chat assistants or general coding agents:
- A library of written scientific procedures, guided workflow templates, catalogs of research data sources, plus reviewer and writer roles the agent can hand work to.
- A Living Lab Notebook whose entries link into an argument and are append-only, never erased.
- Tamper-resistant logging: the system records what happened by watching the agent's actions rather than trusting its claims, in a log the agent has no tool to write to. This targets model overclaiming—the most common failure in their earlier benchmark of nine frontier models—by making claims checkable instead of blocking them.
Released seven months ago, the tool's workflows were shaped by conversations with scientists in academia, government, and industry; the paper also evaluates it on twenty interdisciplinary research prompts under a rubric.
More from coding & agent
- AgentCraft runs a team of Claude coding agents inside Minecraft with real worktrees — max_paperclips · 2026-10-03
- Dev plans human-in-the-loop email gateway so AI agents never touch his inbox directly — tomchapin · 2026-10-03
- Solo dev builds an off-market real estate deal pipeline with Claude — and hits the data wall — italimade · 2026-10-03
- OpenClaw v2026.9.8 ships GPT-6.1 Sol support across 43 PRs, cuts memory use — heyneighbor · 2026-10-03
- Building personal productivity meta-tools is now fun: bash scripts have become full-blown apps — devenbhooshan · 2026-10-03
- Google paper: clean-context verifiers let explorer agents try wild proof ideas safely — solyarisoftware · 2026-10-03