Llmwiki
Open-source, self-hosted LLM wiki that compiles your raw sources into a maintained knowledge base
The Ingest/Query/Lint loop plus an immutable raw/ folder is the right shape for this problem — your sources stay untouched, and good query answers get filed back as pages instead of evaporating into a chat log. It is Apache 2.0 and self-hosted, with a one-click ingest path from the GitHub repo, so the only real cost is setup and API keys. If you need zero-setup hosting or live multi-user editing, this is not the tool; for Git-native researchers, compare it against Obsidian plus an AI plugin or Notion AI, which give you a hosted editor but leave the cross-reference upkeep to you.
Verified 13h ago · liveness 64/100 · cite: rightaichoice.com/tools/llmwiki
- Researchers compiling literature reviews across many papers and PDFs
- Technical writers structuring research notes into a maintained knowledge base
- AI engineers giving agents memory compiled from dynamic source material
- Git-comfortable individuals and teams who want version control and privacy
- Non-technical users who do not want to self-host or manage API keys
- Teams needing real-time collaborative editing with multiple concurrent users
- Anyone wanting a no-code, zero-setup hosted knowledge base
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip LLM Wiki if you want a hosted, zero-setup knowledge base you open in a browser and never administrate — this one asks you to run a server, manage API keys and work in Git.
Running it yourself means paying for model API calls on top of the software, and every ingest that touches 10-15 pages is a multi-page generation, so a large raw-source drop is billed as many calls, not one.
The software is Apache 2.0 and self-hosted, so the licence line is zero and your real budget is model API calls plus a server. That compares favourably with per-seat hosted knowledge bases — a small research team pays for compute instead of seats — but it shifts cost to whoever runs the boxes, and a solo user with light usage may find a hosted AI note tool cheaper once their own time is priced in.
In short
Llmwiki — Open-source, self-hosted LLM wiki that compiles your raw sources into a maintained knowledge base. Best for Researchers compiling literature reviews across many papers and PDFs, Technical writers structuring research notes into a maintained knowledge base, AI engineers giving agents memory compiled from dynamic source material. Free to use.
What people actually say about Llmwiki — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
45 mentions across 5 sources (Hacker News, YouTube, Bluesky, GitHub, Lemmy) · researched Jul 14, 2026.
Average across the 5 sources that answered — each source counts once, not each post.
- +Fully automated wiki maintenance with no manual bookkeeping.
- +Free, open-source, and self-hosted with full data control.
- +Automatically generates summaries, entity pages, and cross-references.
- +Detects contradictions and flags stale claims via health checks.
- +Supports multiple LLM backends via API keys (e.g., Claude).
- −Requires per-project setup, not system-wide knowledge base.
- −Steep learning curve for non-technical users.
- −Small user base means limited community support.
- −Setup process can be confusing, per user reviews.
- −Not a plug-and-play solution; needs manual configuration.
- • LLM API usage costs (e.g., Claude, OpenAI) if self-hosted
- • Self-hosting infrastructure (server, storage, maintenance time)
Viability Score
How well maintained and how widely used is Llmwiki? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- Ingest raw PDF, text, markdown and transcript files into the wiki
- LLM writes summaries for each newly ingested source
- Updates entity and concept pages across the wiki on ingest
- Updates cross-references between pages as knowledge changes
- Flags new material that contradicts existing wiki knowledge
- Ingests a single source into as many as 10-15 wiki pages
- Query the compiled wiki for answers already synthesized from sources
- Files good query answers back into the wiki as new pages
- Lint health checks for inconsistent data and stale claims
- Lint detection of orphan pages and missing cross-references
- Lint suggests new questions to ask and new sources to look for
- Schema config file defines wiki structure, conventions and ingest workflows
- Immutable raw/ source layer the LLM reads but never modifies
- Web interface for searching, querying and linting the wiki
- Recent-changes view of the latest pages, edits and sources
About Llmwiki
LLM Wiki is an Apache 2.0, self-hosted knowledge base where the LLM does the bookkeeping. You drop raw sources — PDFs, markdown files, articles, notes, transcripts — into a raw/ folder, and the model reads them, writes summaries, updates entity and concept pages, and flags anything that contradicts what the wiki already says. The project is presented as a working implementation of Andrej Karpathy's LLM Wiki concept, and it quotes his framing directly: the tedious part of a knowledge base is not the reading or the thinking, it is the bookkeeping. A single ingested source can touch 10 to 15 wiki pages in one pass. The architecture is deliberately three-layered. Raw Sources are immutable — the LLM reads from them but never modifies them. The Wiki layer is LLM-owned markdown: summaries, entity pages, concept pages, cross-references. You read it; the model writes it. The Schema is a config file that tells the LLM how the wiki is structured, what conventions to follow, and what workflows to run on ingest. The homepage demo shows a research wiki with Overview, Concepts, Attention Mechanisms, Scaling Laws, Entities, Transformer Architecture and Sources pages, plus a recent-changes view listing the latest pages, edits and sources. The workflow runs in three stages. Ingest converts a dropped source into summaries plus entity and concept updates across the wiki, and surfaces contradictions against existing knowledge. Query answers questions against the already-synthesized wiki rather than re-deriving answers from raw chunks each time — and good answers get filed back as new pages, so explorations compound. Lint runs health checks for inconsistent data, stale claims, orphan pages and missing cross-references, then suggests new questions to ask and new sources to look for. State is version-controlled through Git. It is built for researchers compiling literature reviews, technical writers structuring research notes, and AI engineers who want agent memory built from dynamic sources — the same compile-once-query-the-synthesis approach that surfaced in a recent Hacker News discussion about avoiding transcript hoarding for session memory. If you already live in Git and self-hosted infrastructure and would rather read a wiki than maintain one, this fits.
Behind the Verdict
The insight LLM Wiki is built on is worth taking seriously: a knowledge base rots not because reading is hard but because keeping twenty pages consistent with one new paper is boring, and boredom is exactly what models do not feel. The three-layer split enforces that discipline structurally. Raw Sources are immutable, so you can always audit what the LLM actually read. The Wiki layer is model-owned markdown, so you get readable plain files rather than a locked database. The Schema file is the clever part — it is where you tell the model your conventions and what workflow to run on ingest, which is what stops a generated wiki from drifting into inconsistency as it grows. The Query stage is where it earns its keep for research work. Because answers are drawn from the synthesized wiki rather than re-chunked raw files, you are not paying to re-derive the same answer every session, and filing good answers back as pages means a literature review gets denser each time you interrogate it. Lint completes the loop with health checks for inconsistent data, stale claims, orphan pages and missing cross-references, plus suggestions for what to ask and what source to go find next. Where it is weak: this is infrastructure, not a product someone hands you. You run it, you supply the API keys, and you need comfort with a server and Git. The writing model is asymmetric — you read the wiki, the model writes it — so it is not a co-editing surface for a team in a browser; concurrent multi-user editing is not what it is designed around. Generated lint output and contradiction flags are model judgments and should be reviewed, not trusted blind. And because one ingest can touch 10 to 15 pages, a large raw-source drop produces broad edits in a single pass — fine when they are right, worth reading the diff when they are not. Git is the safety net there, which is also the reason the tool suits people already fluent in it. Choose it if you are compiling literature, structuring technical research notes, or building agent memory from files that change. Look elsewhere for a hosted, no-setup knowledge base or a team wiki with live co-editing.
Researching Llmwiki? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Llmwiki actually fits — and what changes day-one when you adopt it.
You drop 40 papers into raw/ over a few weeks. Each ingest writes a summary, updates the entity pages for the models and datasets involved, revises concept pages like attention mechanisms and scaling laws, and flags two papers whose reported results contradict what the wiki already says.
Outcome: You end up with a cross-referenced literature wiki you never hand-edited, plus a short list of contradictions to resolve before writing — instead of 40 unread PDFs and a spreadsheet.
You add articles, meeting transcripts and internal notes as they arrive, then query the compiled wiki for 'what have we decided about evaluation methodology' rather than re-reading sources. The answer you like gets filed back as a page.
Outcome: Your notes become a maintained reference with a recent-changes view of every page, edit and source, and Git gives you a diff for each pass.
You point the ingest workflow at dynamic source folders, let Lint surface orphan pages and missing cross-references, and query the synthesized wiki layer instead of stuffing raw chunks into context.
Outcome: Agent memory that is compiled once and queried repeatedly, with the raw sources preserved immutably for audit.
Use Cases
- Compile a literature review from a folder of uploaded papers and PDFs automatically
- Maintain a living entity-relationship wiki for an ongoing research project
- Detect conflicting claims across sources with automated contradiction flags
- Query your knowledge base for synthesized answers instead of re-reading source files
- Run lint health checks to find orphaned or outdated wiki pages
- Give an AI agent memory compiled from dynamic, changing files
- Keep transformer-architecture concepts, entities and sources cross-referenced in one wiki
Models Under the Hood
as of 2026-10-18
Limitations
- LLM Wiki is an Apache 2.0 open-source, self-hosted tool, so setup, hosting and wiki maintenance are on you, and it presumes a server and a Git workflow.
- It follows a three-layer model in which you rarely write the wiki yourself: raw sources are immutable and the wiki layer is LLM-generated, with the LLM writing and you reading rather than a team co-editing in a browser.
- Lint and contradiction flagging output is model-generated, so flagged inconsistencies and stale claims should be reviewed rather than accepted blind.
- A single ingested source may touch 10–15 wiki pages, so large raw-source sets can trigger broad edits per ingest.
as of 2026-10-18
Verification history
We have re-verified Llmwiki 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Llmwiki tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0
Ideal for
Researchers, technical writers and AI engineers willing to self-host and manage their own model API keys in exchange for full control and privacy
What this tier adds
Starting tier — Apache 2.0 software at no licence cost, bundling the Ingest, Query and Lint workflow, the immutable raw/ folder and LLM-owned wiki, the Schema config file, the web interface and Git version control.
Where the pricing makes sense
The company stage and team size where Llmwiki's pricing actually pencils out — and where peers do it cheaper.
The software is Apache 2.0 and self-hosted, so the licence line is zero and your real budget is model API calls plus a server. That compares favourably with per-seat hosted knowledge bases — a small research team pays for compute instead of seats — but it shifts cost to whoever runs the boxes, and a solo user with light usage may find a hosted AI note tool cheaper once their own time is priced in.
Setup time & first value
How long it actually takes to get something useful out of Llmwiki — broken out by persona, not the marketing-page minute.
Expect an afternoon, not five minutes: cloning the repo and running the web interface is quick, but wiring model API keys and writing a Schema file that encodes your conventions is where the real setup time goes. Non-technical users should budget considerably more, or pick a hosted alternative.
Switching to or from Llmwiki
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Obsidian: move your vault notes into raw/ as your immutable source layer, then let Ingest build the LLM-owned wiki layer above them and keep Obsidian for reading.
- →From Notion AI: export pages to markdown, drop them in raw/, and let the Schema file carry over the structure conventions your Notion workspace used.
- →From a plain folder of PDFs: drop the files into raw/ and run Ingest; summaries and entity pages are generated for you rather than written by hand.
- ↗To Obsidian: the wiki layer is markdown, so point an Obsidian vault at it and keep Git as your history.
- ↗To a hosted knowledge base: export the markdown wiki pages and re-upload, accepting that automated cross-reference upkeep and lint checks are what you give up.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Llmwiki”, and we withheld 6: 6 could not be judged, because “Llmwiki” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Llmwiki.
Official links
Tools that pair well with Llmwiki
Common stack mates teams adopt alongside Llmwiki, with the specific reason each pairing earns its keep.
Heptabase
Heptabase: a visual knowledge base with an AI Tutor that turns your sources into structured study sessions.
MyMemo AI
MyMemo AI turns your saved links, videos, PDFs and voice notes into a searchable knowledge base you chat with.
Reflect AI
Open-source, markdown-first note-taking app with daily notes, backlinks, and optional AI on your own API key.
Featured Head-to-Head Comparisons
Llmwiki vs Praktika
If your goal is conversational fluency in a new language, Praktika's AI tutors and real-time corrections are unmatched for intermediate learners. For researchers or writers who need a self-updating knowledge base from raw documents, LLM Wiki is a powerful open-source tool—if you're comfortable self-hosting. Choose based on whether you need speaking practice or document synthesis.
Llmwiki vs Surge Ai
Choose Surge AI if you need rigorous, expert-graded human feedback to improve frontier AI models—especially for complex reasoning or safety testing. Llmwiki is the clear pick for anyone who wants a low-cost, self-hosted wiki that automatically synthesizes and maintains knowledge from raw documents without manual effort.
Alternatives to Llmwiki
View all
Heptabase
Heptabase: a visual knowledge base with an AI Tutor that turns your sources into structured study sessions.
MyMemo AI
MyMemo AI turns your saved links, videos, PDFs and voice notes into a searchable knowledge base you chat with.
Reflect AI
Open-source, markdown-first note-taking app with daily notes, backlinks, and optional AI on your own API key.
Frequently Asked Questions
Best-of guides
Used Llmwiki? Help shape our editorial sentiment research.