Learn Octopus from idea to running app.

Install it, open the local app, run one Goal, and inspect where the tool workflows live.

Start with install

1. Follow the product path

Product

Install -> first-run -> start --open

The local app shows Goal, Need, Feed, Output, and pet state from the run.

Open install guide

2. Install and launch

curl -fsSL https://dangozhang.github.io/Octopus/install.sh | sh
octopus --version

After install, run one Goal and open the local app.

3. Run the first clean loop

octopus first-run "make this repo easier to use"
octopus start --open

Goal

User input changes the objective.

Need

The main loop asks a worker for the next useful result.

Feed

A tentacle supplies evidence and compresses the result.

4. Connect a model

codex login
export OCTOPUS_LLM_BACKEND=codex
export OCTOPUS_LLM_CODEX_COMMAND=codex
octopus provider status
export OCTOPUS_LLM_BACKEND=openai-compatible
export OCTOPUS_LLM_MODEL=gpt-4.1-mini
export OCTOPUS_LLM_BASE_URL=https://api.openai.com/v1
export OCTOPUS_LLM_API_KEY=<api-key>
octopus provider check

provider check verifies the configured endpoint, model, and authentication before live use.

5. Preview field evolution

octopus evolve parallel --workers 1 --open "advance the peer field objectives toward v0.2.0"
octopus fields summary

This opens one worker slot from the peer field pool, runs the editable field harness, records Feed plus verifier signal, and keeps the Octopus pet as a state observer.

6. Find the editable tool workflows

tentacles/

Editable workflows for repo, browser, repair, and script tasks.

crates/octopus-core/

Runtime, state, local server, and release checks.

docs/

GitHub Pages website, tutorial, use guide, recipes, app page, and research map.