Install#

ensō is a single static Go binary (named enso). No CGO, no runtime dependencies on the host beyond a POSIX shell (and optionally podman or lima if you run an isolated backend — see [backend]).

Prerequisites#

  • Go 1.26+ (only if building from source).
  • An OpenAI-compatible chat endpoint — llama.cpp’s llama-server is the reference target; vLLM, Ollama, OpenAI itself, anything that speaks Chat Completions with SSE streaming will work.
  • POSIX OS (Linux/macOS/BSD). Windows builds compile but the daemon path is unsupported there; run via WSL.

Build from source#

git clone https://github.com/TaraTheStar/enso.git
cd enso
make build              # produces ./bin/enso

Common Make targets:

targetwhat it does
make buildCompile to ./bin/enso.
make runBuild + launch the TUI.
make daemonBuild + run the daemon in the foreground.
make testRun the unit test suite.
make checkgofmt + vet + test + build (full pre-commit gate).
make tidyRefresh go.mod / go.sum.
make helpPrint this list.

The Makefile sets CGO_ENABLED=0, -trimpath, and -ldflags '-s -w' by default — the resulting binary is reproducible and stripped.

Setting up a local model#

For the default config, point ensō at llama.cpp’s llama-server. On a single RTX 3090:

llama-server \
  -m unsloth/Qwen3.6-35B-A3B-GGUF/UD-Q4_K_XL.gguf \
  -ngl 999 -c 32768 -fa on --no-mmap \
  -ctk q8_0 -ctv q8_0 \
  --jinja --reasoning-budget 4096 --reasoning-budget-message \
  --temp 0.6 --top-k 20 --top-p 0.95 --min-p 0.0 --presence-penalty 1.5 \
  --port 8080

Any other backend works — set endpoint and model in ~/.config/enso/config.toml (or any other layered config path) to point elsewhere. See the config reference for the full list.

Optional: isolated backend#

If you want the agent to run inside a per-project container, install one of:

  • podman (rootless, no daemon) — recommended.
  • docker — works equally well; runs as the docker daemon’s user.

Then set [backend] type = "podman" in your config. ensō’s container runtime defaults to auto-detect (podman, falling back to docker); pin one with [backend.podman] runtime = "podman" or "docker". (For a VM backend instead, use [backend] type = "lima".) Details in the sandbox page.

Optional: language servers#

Configure [lsp.<name>] blocks for each language you work with. Common servers:

  • gopls — Go.
  • rust-analyzer — Rust.
  • typescript-language-server --stdio — TypeScript / JavaScript.
  • pyright-langserver --stdio — Python.
  • clangd — C / C++.

See LSP integration for example configs. Servers are spawned lazily on first use and reused for the session.

Next: Quickstart#