Swarmvault
Local-first LLM wiki, knowledge graph & RAG knowledge base with CLI and desktop app.
Swarmvault delivers a genuinely durable, local-first knowledge graph with provenance, schema-guided compilation, and anti-drift linting—at zero cost and without API keys. It stands out for agent-heavy workflows via MCP exposure to Claude Code, Cursor, and Codex. The CLI learning curve is steep; non-technical users should prefer NotebookLM or Obsidian. For developers and researchers, it's a standout free option.
Verified 1d ago · liveness 60/100 · cite: rightaichoice.com/tools/swarmvault
- Developers building a local-first knowledge graph with provenance tracking
- Researchers compiling papers (arXiv, DOI) into a durable, queryable wiki
- Knowledge workers wanting a private, offline RAG knowledge base on their own machine
- Users of Claude Code, Cursor, or Codex needing a persistent agent memory layer via MCP
- Non-technical users preferring a GUI-only, zero-config tool
- Teams needing real-time collaboration or cloud sync out of the box
- People who want a simple chat-with-docs tool without a persistent wiki graph
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 Swarmvault if you need a zero-config GUI, real-time collaboration, cloud sync, mobile access, or a hosted solution—it's CLI-first and local-only.
Optional API keys for OpenAI, Anthropic, Gemini, or Ollama incur their own usage costs when you enable them for richer synthesis.
Swarmvault is free and open-source (MIT), making it the cheapest option for developers and researchers. It competes with paid tools like NotebookLM (free tier exists but limited) and Obsidian (free but RAG plugins may cost). For teams needing collaboration, paid alternatives like Notion AI or Confluence may be better despite cost.
In short
Swarmvault — Local-first LLM wiki, knowledge graph & RAG knowledge base with CLI and desktop app. Best for Developers building a local-first knowledge graph with provenance tracking, Researchers compiling papers (arXiv, DOI) into a durable, queryable wiki, Knowledge workers wanting a private, offline RAG knowledge base on their own machine. Free to use.
What's new in Swarmvault
Checked yesterdayAcross the latest 4 updates: 4 feature updates.
Agent context packs and task ledger added
Introduced 'swarmvault context build' for bounded context packs and 'swarmvault task' ledger for recording agent task goals, decisions, and outcomes.
Guided ingestion sessions with --guide flag
New '--guide' flag for 'swarmvault ingest' and 'source add' generates resumable source sessions and reviews before committing updates.
Token-budgeted compile and git-aware commits
Added 'compile --max-tokens' to trim generated pages and '--commit' for auto-committing changes to git.
Agent-ready auto-install rules for Claude, Cursor, Codex
Auto-install rules for Claude Code, Cursor, and Codex enable directly exposing vault via MCP to compatible agents.
Viability Score
How well maintained and how widely used is Swarmvault? 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: September 2026
How we score →Key Features
- Local-first RAG knowledge base with provenance
- CLI pipeline: init, ingest, compile, query, tasks, MCP, graph serve
- Schema-guided wiki via swarmvault.schema.md
- Anti-drift linting for stale sources, conflicts, missing citations
- Agent context packs via 'swarmvault context build'
- Task ledger via 'swarmvault task'
- Guided ingestion sessions with --guide flag
- Token-budgeted compile with --max-tokens
- Git-aware commits with --commit
- MCP exposure for Claude Code, Cursor, Codex
- Graph report pages with god nodes and communities
- Optional model providers: OpenAI, Anthropic, Gemini, Ollama
- Content-hashed immutable source manifests
- Desktop app reads local graph.json, report.json, approvals, candidates
- Hybrid search with rerank when embedding provider enabled
About Swarmvault
Swarmvault is an open-source, self-hosted knowledge graph builder and RAG knowledge base that turns raw sources—code repos, arXiv papers, books, slide decks, screenshots, URLs—into a persistent, provenance-tracked wiki. Built on Andrej Karpathy's LLM Wiki pattern, it runs entirely on your machine. Start fully offline with the built-in heuristic workflow; no API keys required. Optionally connect OpenAI, Anthropic, Gemini, Ollama, or any OpenAI-compatible endpoint for richer synthesis. Everything is CLI-first—init, ingest, compile, query, tasks, retrieval, explore, lint, watch, MCP, and graph serve give you fine control. A companion desktop app reads the same local files (graph.json, report.json, approvals, candidates) so you can visualize the graph, review queues, and search without leaving your machine. The schema-guided approach uses a swarmvault.schema.md file to enforce naming rules, categories, and grounding expectations, while anti-drift linting catches stale sources, conflicting claims, and missing citations. Recent releases add agent context packs via 'swarmvault context build', a task ledger via 'swarmvault task', guided ingestion sessions with a '--guide' flag, token-budgeted compile with '--max-tokens', and git-aware commits with '--commit'. Auto-install rules for Claude Code, Cursor, and Codex let you expose the vault to AI agents over MCP, making it a practical memory layer for agent workflows. Swarmvault is for developers, researchers, and tech-savvy knowledge workers who want a durable second brain that compounds over time. Unlike ephemeral chat-with-docs SaaS, it prioritizes longevity and ownership—everything stays as local files. For non-technical users, NotebookLM or Obsidian with a RAG plugin might be easier, but if you value depth, provenance, and agent integration, Swarmvault is a compelling free, open-source choice.
Behind the Verdict
Swarmvault is more than a chat-with-docs tool: it builds a persistent wiki that compounds. The three-layer architecture (raw sources, wiki, schema) follows Karpathy's LLM Wiki pattern, and the local-first design means you own everything—no cloud, no lock-in. The CLI is powerful but has a steep learning curve for newcomers; however, the desktop app eases graph visualization and review. Anti-drift linting is a differentiator, catching stale sources and conflicting claims automatically. The optional model providers (OpenAI, Anthropic, Gemini, Ollama, or any OpenAI-compatible) add richer synthesis when needed, but the heuristic offline path works without them. For teams needing collaboration or cloud sync, it's not suitable; there's no real-time multi-user support. For developers building agent memory layers, the MCP auto-install rules for Claude, Cursor, and Codex are compelling. Overall, if you value provenance, longevity, and open-source ownership, Swarmvault is a strong choice—but it's not for everyone.
Researching Swarmvault? 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 Swarmvault actually fits — and what changes day-one when you adopt it.
Install the CLI, run 'swarmvault quickstart /path/to/repo', then 'swarmvault compile' to generate a wiki with module pages.
Outcome: Get a searchable wiki of the codebase with provenance, ready for future reference and agent queries via MCP.
Use 'swarmvault add' with arXiv URLs, then run compile to create source pages with normalized metadata.
Outcome: Build a durable, queryable repository of research papers with cross-references and citation tracking.
Run 'swarmvault mcp install' for Claude Code, then 'swarmvault context build' for a specific task.
Outcome: Expose the vault to agents, giving them a persistent memory layer with bounded context packs for PR reviews or handoffs.
Use Cases
- Compile a personal knowledge graph from research papers, notes, and code repositories with provenance tracking.
- Set up a local RAG knowledge base that persists answers and builds a wiki over time, usable offline.
- Create a durable memory store for Claude Code, Cursor, or OpenClaw agents to retain context across sessions.
- Lint your knowledge base for stale sources, conflicting claims, and coverage gaps automatically.
- Expose your vault to MCP-compatible clients for agent-driven exploration and query.
- Build a fan wiki chapter by chapter (e.g., for books) with character, theme, and plot pages that compound.
- Use guided ingestion sessions to review and integrate sources incrementally.
- Maintain a personal memex with journals, health logs, and goals cross-referenced into a structured wiki.
Models Under the Hood
as of 2026-09-01
Limitations
- Swarmvault is a local-first tool storing all data as local files (raw sources, markdown wiki, and graph state).
- It can work offline with a built-in heuristic path, while external model providers (e.g., OpenAI, Anthropic, Gemini, Ollama) are optional upgrades for richer synthesis.
- It is CLI-first with a desktop app, and non-technical users may find the CLI learning curve steep.
as of 2026-09-01
Verification history
We have re-verified Swarmvault 7 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-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
Showing the 6 most recent of 7 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 Swarmvault 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
Developers, researchers, and tech-savvy knowledge workers who want a free, self-hosted knowledge graph with full control and provenance.
What this tier adds
Free, MIT-licensed, includes all CLI tools, desktop app, and local-first operation; no paid tiers exist.
Where the pricing makes sense
The company stage and team size where Swarmvault's pricing actually pencils out — and where peers do it cheaper.
Swarmvault is free and open-source (MIT), making it the cheapest option for developers and researchers. It competes with paid tools like NotebookLM (free tier exists but limited) and Obsidian (free but RAG plugins may cost). For teams needing collaboration, paid alternatives like Notion AI or Confluence may be better despite cost.
Setup time & first value
How long it actually takes to get something useful out of Swarmvault — broken out by persona, not the marketing-page minute.
For a developer, the initial install (npm install -g @swarmvaultai/cli) and quickstart can be done in under 15 minutes. For a researcher, adding sources and compiling a small vault may take 30-60 minutes. Non-technical users may need a few hours to get comfortable with the CLI.
Switching to or from Swarmvault
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From Obsidian: import existing markdown notes into raw/ and let Swarmvault compile them into a structured wiki.
- →From Notion: export pages as markdown and add them to raw/ for compilation.
- →From Evernote: export notes as HTML or markdown and ingest into Swarmvault.
- ↗To Obsidian: Swarmvault's wiki/ is pure markdown, so you can open it directly in Obsidian.
- ↗To Notion: import the markdown files into Notion for cloud collaboration.
- ↗To standard markdown: use 'swarmvault export ai' to get llms.txt and full text for other tools.
Integrations
Resources & Guides
Tutorials & Learning
Official links
Tools that pair well with Swarmvault
Common stack mates teams adopt alongside Swarmvault, with the specific reason each pairing earns its keep.
GraphRAG
Open-source knowledge-graph RAG that maps entities and communities to answer complex, cross-document questions.
Heptabase
Visual knowledge base with AI Tutor for guided deep learning and research
Obsidian Smart Connections
Local-first AI plugin that surfaces related notes in Obsidian with no API key required.
Featured Head-to-Head Comparisons
Swarmvault vs Surge Ai
Choose Swarmvault if you're a developer wanting a free, private knowledge graph with offline RAG. Choose Surge AI if you need expert human feedback for training or evaluating frontier models, backed by new rigorous benchmarks and enterprise trust.
Swarmvault vs Praktika
Choose Praktika if you want on-demand speaking practice with AI tutors; choose Swarmvault if you need a private, local-first knowledge graph for research and development. They serve entirely different needs—language learning vs. personal wiki/RAG. No news overrides static facts.
Alternatives to Swarmvault
View allGraphRAG
Open-source knowledge-graph RAG that maps entities and communities to answer complex, cross-document questions.
Obsidian Smart Connections
Local-first AI plugin that surfaces related notes in Obsidian with no API key required.
Frequently Asked Questions
Best-of guides
Used Swarmvault? Help shape our editorial sentiment research.