Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|---|---|---|
| .beads | ||
| 9p@439ee6af1e | ||
| agents | ||
| core@b7c1b317cb | ||
| doc | ||
| el@7b6bb80d53 | ||
| prompts | ||
| sandbox | ||
| scripts | ||
| skills@2dd6fea2e6 | ||
| tools | ||
| tui@bad5ad4fda | ||
| .gitignore | ||
| .gitmodules | ||
| Containerfile | ||
| README.md | ||
| mkfile | ||
README.md
ollie
Ollie starts from a simple, slightly silly question: What might an AI agent look like in Plan 9? From there follow two more: What happens if agent primitives are exposed as an ordinary filesystem? and How much can we subtract from agent implementations while still being useful? The result is not a conventional "agent framework" but a small Go runtime and 9P‑backed filesystem surface that defines what an agent is, then exposes its state and behaviors as regular files and I/O streams, so that the surrounding environment (shell scripts, TUIs, web apps, cron, containers, and other OS facilities) can handle orchestration, scheduling, workflows, and interfaces instead. Features like conductor agents, complex workflow engines, or fancy editors are optional layers you build yourself by composing ordinary file reads, writes, and processes: a tiny s/sh CLI, an optimize | b/q pipeline, or a Claude‑Code‑style UI talking to the same filesystem namespace.
Getting Started
Clone with submodules:
git clone --recurse-submodules https://github.com/lneely/ollie.git
Or after cloning:
git submodule update --init --recursive
Build and install everything from the monorepo root:
mk
Or build specific components:
mk core # Build core library
mk tui # Build terminal UI
mk 9p # Build 9p filesystem server
For Emacs Lisp, copy el/ellie.el into your own Emacs configuration directory, and:
(require 'ellie)
See doc/USAGE.md for usage instructions.
Configuration
Environment: ~/.config/ollie/env
OLLIE_BACKEND=openai # ollama | openai | openrouter | anthropic | copilot | kiro (default: ollama)
OLLIE_OLLAMA_URL= # base URL for Ollama (default: http://localhost:11434)
OLLIE_OPENAI_URL=https://openrouter.ai/api
OLLIE_OPENAI_KEY=sk-or-...
OLLIE_ANTHROPIC_KEY=sk-ant-...
OLLIE_COPILOT_TOKEN=...
OLLIE_KIRO_TOKEN=... # bearer token or sqlite:// path (auto-detected from Kiro CLI if unset)
OLLIE_MODEL=qwen/qwen3-235b-a22b
OLLIE_TOOLS_PATH=~/.config/ollie/tools # directory for tool scripts ({tool} steps)
OLLIE_MEMORY_PATH=~/.config/ollie/memory # directory for memory files (ollie/m)
OLLIE_BEADS_ENABLE=0 # set to 0 to use markdown-based plan_* tools instead of beads
Shell environment variables take precedence over the env file.
Remote Access
Because olliesrv speaks 9P over TCP, a remote instance can be mounted into the local namespace using 9pfuse. This means agent sessions, tools, and all other filesystem state on a remote host are accessible as ordinary files — no special client needed.
Start a remote server
On the remote host (or in a container):
olliesrv start -tcp :9564
Mount the remote namespace locally
olliesrv mount remotehost:9564 ~/mnt/remotehost
ls ~/mnt/remotehost # s/, t/, b/, ...
Sessions created under ~/mnt/remotehost/s/ run on the remote host, so tool calls execute close to the remote filesystem rather than over the wire.
Container
A Containerfile is included for running a self-contained remote server:
podman build --network=host -t olliesrv .
podman run --network=host -e ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY olliesrv
Then mount locally:
olliesrv mount localhost:9564 ~/mnt/container-ollie
Development
Update all submodules to latest:
git submodule update --remote
Credits
Many sources of inspiration:
- Plan 9 from Bell Labs — for an interesting system
- @9fans — for the Plan 9 port
- Suckless — for articulating good software development principles
- @simonfxr — for a solid agent baseline to "borrow" from, and other nifty ideas
- @aws — for a solid open-source agent implementation
License
GPLv3