A local server that gives any AI coding tool a persistent memory, a map of your code, cheap delegate models, and guardrails it can't write past. Point your tool at it, or run it alongside. Your context follows you between tools.
Two parts:
- aimee-server is an assistant to one human. Learns how you work and what you expect. General purpose, but coding is what it does best, especially across large or multi-repo work.
- aimee-kb is a knowledge base for a whole corpus: a subject, a company, a team.
- Memory across sessions and tools. A curator distills each session into a typed knowledge base, dedupes facts, flags contradictions, and lets stale detail decay. Point a team at one kb and everyone works from the same memory. See KNOWLEDGE.md.
- Your code as a graph. aimee indexes your code into a symbol and call graph, spanning repos. The AI finds callers and traces an edit's blast radius before it writes. See CODE_INTELLIGENCE.md.
- Delegates that cut the bill. Route review, summarization, and boilerplate to the cheapest model that can do it. A local GPU model or a subscription plan costs nothing extra. The primary gets a compact result back. See DELEGATES.md.
- Guardrails.
.env, keys, and prod configs are blocked before the AI touches them. Planning mode freezes writes. Every session is fully isolated by aimee. See SECURITY.md. - Any model, any provider. Point a tool's OpenAI or Anthropic compatible API at aimee and it runs the turn on any model: Claude, GPT, Gemini, a model on your own GPU. Or run aimee beside the tool over MCP/ACP and it keeps its own model. Switch tools whenever.
- Workflows. Compose a job from typed steps and aimee runs it to a PR, with review panels and human sign-off gates where you want them. See WORKFLOWS.md.
- Run your own inference. Embeddings, reranking, and synthesis in one CPU or GPU container. The kb curates locally, no outside API calls. See KB_LLM_BACKENDS.md.
- A browser workspace.
aimee-webchat: chat, a live code graph, a git project manager, an in-app VS Code editor, and dashboards. No terminal required. See DASHBOARD.md.
Core services are C. Hot paths run in single-digit milliseconds. Nothing phones home.
Run the services in Docker, install the aimee CLI on each machine.
# Run the stack (server, kb, Postgres, embedder)
git clone https://github.com/RakuenSoftware/aimee.git
cd aimee
docker compose -f compose.combined.yaml up --build -d
# Confirm it's live (default bearer: aimee-local-dev; -k for the self-signed cert)
curl -k -H 'Authorization: Bearer aimee-local-dev' https://localhost:8743/v1/healthThen point the client at the server and register your tool's hooks:
aimee remote set https://host:8743 <token> # set AIMEE_TLS_INSECURE=1 for the self-signed cert
./configure-hooks.shQUICKSTART.md walks the full setup. Every deployment topology and the from-source install live in Deployment and operations.
aimee memory store myhost "PVE at 10.0.0.1" # a fact the AI keeps across sessions
aimee memory search "proxmox"
aimee delegate review "Review this PR" # route to a cheaper model
aimee status| Document | What's in it |
|---|---|
| Manual | Full reference: install, config, commands, feature guides. |
| Architecture | Processes, storage, trust boundaries, request lifecycles. |
| Technical Reference | Code internals, RPC/HTTP surfaces, build, deployment. |
| Commands | Client command contract, flags, options. |
| How aimee learns | The kb, the self-learning pipeline, delegation economics. |
| Code intelligence | Symbol/call graph, cross-repo deps, blast radius. |
| Workflows | The dev-lifecycle workflow engine and authoring. |
| Security Model | Threat model, trust boundaries, capability system. |
| Compatibility | Supported OS, shell, and provider matrix. |
| Feature Status | Implementation status of all features. |
More references live under docs/.
Questions and discussion on the aimee Discord: https://discord.gg/FjGjvcgAqz.
Copyright (C) 2026 The aimee authors. Licensed under the GNU AGPL v3.0 (AGPL-3.0). See LICENSE and NOTICE.
If the AGPL doesn't suit you, other terms can be discussed. Contact jbailes@gmail.com.
Bundled third-party components (cJSON, the generated SDKs under api/sdks/) are licensed
separately; see NOTICE.