One shared, durable memory for your AI coding agents — local-first, zero-dependency.
Model Context Protocol (MCP) Server: io.github.VonderVuflya/yggdrasil
The io.github.VonderVuflya/yggdrasil MCP server provides a shared, durable memory for AI coding agents. It is described as local-first and zero-dependency, supporting reuse across sessions, tools, and projects. The excerpt specifically mentions Claude Code, Codex, and “every MCP agent,” with data kept on the machine.
🛠️ Key Features
Shared memory for AI coding agents
Durable and local-first
Zero-dependency
🚀 Use Cases
Stop re-explaining a project across AI sessions
Use with Claude Code, Codex, and other MCP agents
Provide context for coding-assistant workflows
⚡ Developer Benefits
Reuse memory across sessions, tools, and projects
Local handling: “Nothing leaves your machine”
Supports MCP-based setups
⚠️ Limitations
Available info does not specify configuration options, API shape, or tool count.
Stop re-explaining your project to every new AI session.
One local memory for Claude Code, Codex, and every MCP agent — shared across sessions, tools, and projects. Zero dependencies. Nothing leaves your machine.
Yggdrasil — a brand-new session already knows your project, and recalls a fix from another project
Every new chat, your AI forgets. You re-explain the project, the decisions, the gotchas — every time, in every tool. Yggdrasil is a tiny always-on memory that any agent plugs into. Open a new session, in any project, with any AI, and it already knows what you decided, what broke, and what's still open.
text
$ cd ~/projects/checkout-api && claude # a brand-new session
🌳 Yggdrasil (injected automatically at session start)
• [project_status] payments refactor: idempotency keys added; open: e2e tests
• [lesson] webhook 401 → signing secret rotated; update env + redeploy
> "have I solved a flaky websocket reconnect anywhere before?"
🌳 recall → found in project `realtime-dash`:
refresh the token *before* opening the socket, then retry with capped backoff.
No "let me remind you what we did yesterday." It's just there.
🚀 Install
Two commands, inside Claude Code (the plugin launches via uv):
The engine lazy-starts on first use and generates its own local token — no API key, no cloud, nothing to configure. Codex and Cursor use the same flow.
All other channels — CLI daemon, Homebrew, npm, Claude Desktop, from source…
drag the .mcpb from the latest release onto Settings → Extensions, paste your token (ygg token) — the desktop app then shares the same memory as your CLI agents (guide)
ygg install is a one-time guided setup: it installs an always-on background service, registers the MCP tools with every agent host it finds — Claude Code, Codex, OpenCode — and, if your hardware allows, recommends optional local models (or pick none to stay zero-config).
OpenCode — nothing to configure
Install OpenCode first, then run ygg install (or ygg redeploy if Yggdrasil is already set up) — the entry is written for you and merged into any existing opencode.json. Confirm with:
bash
opencode mcp list # -> ✓ yggdrasil connected
Installed OpenCode after Yggdrasil? Just re-run ygg install.
If you'd rather write it by hand, note that OpenCode's schema differs from Claude's in four places at once — servers live under mcp (not mcpServers), type is required, command is one array (not command + args), and env is environment (not env), so the Claude snippet won't port:
No token goes in the config — the engine reads the 0600 ~/.yggdrasil/token itself. Run ygg doctor if the tools don't show up.
There is also a yggdrasil-memory skill for any Claude surface: MCP connects the tools, the skill teaches the agent when to use them. Use both for the best behavior.
Try it with nothing installed and a throwaway DB: uvx --from yggdrasil-memory ygg serve --reset --db /tmp/ygg.sqlite.
Then just work: ask your agent "recall what we decided about this project", tell it "remember this decision" — next session it's already there. Verify the install any time with ygg doctor.
Already have history? Seed memory from your existing Claude Code + Codex transcripts, Obsidian vaults, and CLAUDE.md repos — distilled locally:
bash
ygg seed --dry-run # see what it would import; drop the flag to distill for real
Leaving another memory tool?ygg import --from mcp-memory --path memory.json pulls its whole store into Yggdrasil (deduped, secret-guarded) — then you can delete it.
Why
🧠 Persistent — decisions, lessons, and project status survive across sessions.
🔌 One brain, every tool — Claude Code, Codex, OpenCode, and any MCP host share the same memory.
🌐 Cross-project recall — "this looks like what you did in project B — reuse it?"
🧹 Curated, not captured — your agent saves the few things that matter; governance dedupes and archives, never deletes.
🌱 Self-maintaining(opt-in) — a small local model consolidates memory in the background. Zero API tokens.
🪪 One identity everywhere — an optional name and persona every agent picks up, so Claude Code and Codex feel like the same assistant.
🔒 100% local — your memory lives on your machine. No cloud, no account, no telemetry.
🧠 How it works
Yggdrasil is memory + tools — the intelligence is your LLM. It just makes sure the right memory is in front of the right agent at the right moment.
🛎️ Always-on daemon — a tiny local service (~21 MB RAM) your agents reach over MCP tools (ygg_search, ygg_recall, ygg_remember …).
🪝 Hooks — session start auto-injects identity, project status, and open follow-ups (~300 tokens); an optional per-prompt hook auto-recalls memory relevant to each request.
📌 Ranking — pinned and frequently-recalled memories surface first.
🧹 Governance — duplicates and conflicts are queued for review; changes are non-destructive (archive, never delete).
📓 Obsidian — every memory doubles as a plain-Markdown note you can read, edit, and grep.
🎛️ Memory tiers — zero-config by default
Out of the box, Yggdrasil runs on SQLite + FTS5 with zero dependencies — instant keyword search, no models, nothing to download. Optional local models add two independent tiers:
Tier
You add
You gain
0 · default
nothing — SQLite + FTS5
keyword search, zero deps, instant — recall@1 = 0.77
1 · semantic
an embedding model (all-minilm 45 MB · paraphrase-multilingual ~560 MB)
search by meaning, across languages — recall@1 = 0.94, recall@3 1.00
2 · self-maintaining
a small LLM (qwen2.5:1.5b ~1 GB)
background dedupe/merge of memory (propose-only)
The runtime only computes vectors and runs the background model — every memory and every vector stays in the same local SQLite.
ygg install scans for Ollama, LM Studio and llama.cpp, offers to start whichever is installed but idle, and then shows one menu per job with green for models you already have and red for models it would download. Pick a row, and it writes embed_backend, embed_url and distill_url for you — including the part nobody guesses right, that embed_url wants the /v1 base while distill_url wants the host root. A runtime on another machine is one URL: paste it and the dialect is detected. ygg recommend shows the same scan plus the full catalog without changing anything.
Full model menu
Embeddings (semantic search):
Model
Size
Good for
all-minilm
45 MB
English, tiny & fast
nomic-embed-text
274 MB
English, better quality (768d)
mxbai-embed-large
670 MB
English, high quality (1024d)
paraphrase-multilingual
~560 MB
multilingual (EN/RU + 50 langs, 768d)
bge-m3
1.2 GB
multilingual, top quality (heavier)
Embedding backend — Ollama by default. ygg install sets all of this for you
once you pick a runtime; the manual route below is for scripted setups and for
changing one thing later. To use an OpenAI-compatible /v1/embeddings server
instead (llama.cpp's llama-server --embeddings, OpenRouter, LM Studio, vLLM),
set embed_backend:
bash
# local llama.cpp — no key needed
ygg config set embed_backend openai
ygg config set embed_url http://127.0.0.1:8080/v1
ygg config set embed_model bge-small-en-v1.5
ygg redeploy
# OpenRouter — free embeddings, no GPU needed
ygg config set embed_backend openai
ygg config set embed_url https://openrouter.ai/api/v1
ygg config set embed_model nvidia/llama-nemotron-embed-vl-1b-v2:free
ygg config set embed_api_key sk-or-... # or export YGG_EMBED_API_KEY
ygg redeploy
The key is stored in ~/.yggdrasil/embed_api_key (0600) rather than
config.json, and reaches the daemon as a file path — so it never shows up
in ps, the launchd plist or the systemd unit. ygg config list masks it.
Check it took with ygg doctor — dense should name your model:
code
✓ dense active (nvidia/llama-nemotron-embed-vl-1b-v2:free)
LM Studio: the four things that catch people out
ygg install handles all of this. Read on only if you're wiring it by hand.
1. Two settings, two different shapes of the same URL.embed_url is the
/v1 base; distill_url is the host root. Same server, and swapping them gets
you a 404 that reads like the endpoint is simply wrong.
bash
ygg config set embed_backend openai
ygg config set embed_url http://127.0.0.1:1234/v1
ygg config set distill_url http://127.0.0.1:1234
2. The model id is not what you downloaded.lms get nomic-embed-text puts
a model on disk that the API answers to as
text-embedding-nomic-embed-text-v1.5. Ask the server, don't guess:
3. Turn on Just-In-Time model loading (Developer tab). Without it nothing is
loaded when the daemon calls, and every request 404s.
4. Turn on "run the server on login". The Yggdrasil daemon starts at boot; if
LM Studio's server doesn't, dense search silently degrades to lexical until you
next open the app.
Note that GET /api/v1/models answers 200 OK for any key, valid or not —
it ignores auth entirely, so it can't tell you whether your key works. Check
GET /api/v1/key instead: it returns is_provisioning_key, and fails outright
on a bad key.
2. Privacy settings silently hide most models. If a model 404s with
All providers have been ignored, the model is fine — your account is
filtering out every provider that serves it. Fix it at
openrouter.ai/settings/privacy.
That filter is also why openai/text-embedding-3-* can come back 403 on a
provider's terms of service.
Staying local still wins on quality and privacy: on the 232-memory / 110-query
corpus, local paraphrase-multilingual scores recall@1 0.964 vs 0.946
for the free hosted model — and your memories never leave the machine. Reach for
a hosted backend when the box can't run Ollama, not to chase accuracy.
Bigger vectors do not buy accuracy here — on the same corpus
mxbai-embed-large (1024d) scores 0.809 and nomic-embed-text (768d) 0.818,
a difference their confidence intervals swallow whole. What actually moves the
number is whether the model handles your languages: both are English-only and
collapse to 0.40–0.45 on cross-language queries, where the multilingual default
holds 0.95.
Background consolidation (small LLM):
Model
Size
Good for
qwen2.5:0.5b
~400 MB
tiny, fast on CPU
qwen2.5:1.5b
~1 GB
best CPU default
llama3.2:3b
~2 GB
better quality, slower on CPU
The engine itself is swappable — any service meeting the MemoryBackend contract is a drop-in (YGG_ENGINE_URL); see docs/backend-boundary.md.
📊 The numbers
Measured by eval/ygg_eval.py — 232 memories, 110 labelled queries, ranking weights tuned on the dev split only, so holdout is the unbiased number (recall@1, with the paraphrase-multilingual model):
Search view
holdout recall@1
recall@3
zero-dep lexical
Within a project (the real path, pool ~11)
0.94
1.00
0.76
Whole store (no filter, pool 232)
0.72
0.87
0.69
Within a project — the path you use — the right memory is #1 for 0.94 of queries and in the top 3 every time (recall@3 = 1.00). Searching the whole store with no filter is harder (recall@1 0.72, recall@3 0.87 across all 232). Zero-dep lexical mode already solves keyword and code-identifier queries (1.00); the local model adds meaning and cross-language (crosslingual 0.25 → 0.95). The full breakdown in BENCHMARKS.md has 95% CIs, pool sizes, and per-class scores — rerun it in a minute: python3 eval/ygg_eval.py --report.
🆚 Yggdrasil vs the rest
Everyone else either auto-captures transcripts or sells you a cloud. Yggdrasil's bet: keep the few things that matter, curated and de-duped, in plain rows you own — and share them across every tool and project.
Curated decisions / lessons / status (not transcripts)
✅
⚠️ auto-notes
❌ captures everything
⚠️
⚠️ free-form notes
One memory across tools
✅
❌ vendor-siloed
✅
✅
✅
Cross-project recall ("solved this in project B")
✅
❌ repo-scoped
⚠️
⚠️
⚠️
100% local by default
✅
✅
⚠️ cloud sync add-on
❌ hosted-first
✅
Zero dependencies (stdlib + SQLite)
✅
—
❌ Node + Bun + worker daemon
❌ Docker + Qdrant + LLM key
❌
Works with no LLM & no API key
✅
✅
❌ AI-compresses
❌
✅
Semantic search, fully local
✅ opt-in Ollama
❌ grep-only
⚠️ optional Chroma
⚠️ needs API key or Docker stack
❌
Plain Markdown you own (Obsidian-ready)
✅
✅
❌
❌
✅
Closest neighbor — claude-mem: capture-everything memory that records and AI-compresses every session (Node 20+ and Bun, a persistent worker daemon; Chroma optional). Yggdrasil is the opposite bet: a small, high-signal store instead of a growing firehose. mem0 is an SDK plus a hosted platform for building apps that remember their users — even self-hosted it needs an LLM API key. Built-in memories are genuinely useful — and structurally siloed: one vendor, one repo, one machine, literal grep. Yggdrasil is the layer above them (and ygg seed can bootstrap itself from those same transcripts). Different layer entirely: context-mode (live context window) and Context7 (fresh library docs) — both pair fine with Yggdrasil.
🧰 Commands
Agents see six MCP tools: ygg_health, ygg_bootstrap, ygg_search, ygg_recall, ygg_remember, ygg_materialize — auto-registered by the plugin or ygg install.
Full ygg CLI reference
Memory ops
Command
What it does
ygg recall --query "…"
Cross-project search — "have I done this anywhere?"
Guided setup · diagnose with actionable fixes · upgrade
ygg config
Show/set persistent settings (list · get · set · unset)
ygg status · start · stop · restart · logs
Manage the always-on daemon
ygg hooks · unhooks · register
SessionStart hook on/off · (re)register MCP
ygg recommend · token · uninstall
Model catalog · print auth token · remove everything
Give it a personality — edit ~/.yggdrasil/identity.json:
json
{"name":"Jarvis","persona":"concise, proactive, dry wit","user_facts":["prefers TypeScript","ships small PRs"]}
Heavy seeding, weak laptop? Point distillation at any box on your LAN — a desktop with Ollama, LM Studio, llama.cpp, even an iPhone running a local-LLM server app: ygg config set distill_url http://<box>:11434. Yggdrasil auto-detects the API dialect (Ollama or OpenAI-compatible); your data still never leaves your network — details in docs/ygg-cli.md.
❓ FAQ
Claude Code already has built-in memory — why Yggdrasil?
Built-in memories are per-vendor, per-repo, per-machine, and retrieved by literal text match. Yggdrasil is the layer above: the same memory in Claude Code, Codex, and any MCP host, recall across projects, optional semantic search — still 100% local. It bridges them both ways: ygg seed distills your existing native memory + transcripts into the shared brain, and ygg export-native writes a curated digest back into AGENTS.md/MEMORY.md — so even a fresh clone or a tool without Yggdrasil still gets your curated memory.
Does it send my code or memory to the cloud?
No. The engine, the database, and the optional models all run locally. No account, no telemetry. The only outbound call is a version check against PyPI.
Does it automatically remember everything?
No — by design. Retrieval is automatic; writing is deliberate (the agent calls ygg_remember for durable lessons). Capture-everything pollutes memory and burns tokens, so we don't. The optional background model consolidates what's already saved (propose-only).
Do I need a GPU or an API key?
No. The default is pure lexical search — zero dependencies, instant. Semantic search is opt-in and uses a local model via Ollama. The installer recommends one that fits your hardware.
How heavy is it, and what does it cost in tokens?
The engine idles at ~21 MB RAM (lexical default) with ~0% CPU; disk is tens of KB per memory. Session start injects ~300 tokens; each tool call returns a small snippet. All heavy work (indexing, embeddings, consolidation) runs off-LLM on your machine.
Can I edit or delete memories by hand?
Yes. Memories materialize to Markdown notes in an Obsidian vault — read, edit, or remove them like any file. The engine never hard-deletes; it archives (reversible).
🚦 Status & roadmap
Alpha. The happy path and the governance loop are gate-tested (scripts/run_gates.sh); not yet hardened for multi-user or production use. macOS today; Linux/Windows service installers are built and in final on-device testing.
Issues and PRs welcome. Run scripts/run_gates.sh and python3 -m unittest discover -s tests before submitting — all gates must stay green.
📜 License
GNU AGPL v3.0 — see LICENSE. Free and open source: use, modify, self-host, redistribute. If you modify it or offer it as a network service, you must release your source under the same license.
Install
Configuration
Environment variables
YGG_ENGINE_URLdefault http://127.0.0.1:42069
Base URL of the Yggdrasil engine HTTP API.
YGG_ENGINE_TOKENsecret
Bearer token for authenticating to the Yggdrasil engine (from ~/.yggdrasil/token after `ygg install`).
