Spike belongs in your terminal, too.

Install Spike Agent

curl -fsSL https://spikeai.de/spike/cli/install.sh | sh && export PATH="$HOME/.local/bin:$PATH"

powershell.exe -NoProfile -ExecutionPolicy Bypass -Command "irm https://spikeai.de/spike/cli/install.ps1 | iex"

curl -fsSL https://spikeai.de/spike/cli/install.sh | sh && export PATH="$HOME/.local/bin:$PATH"

Then check the installed command:

spike --version

Sign into your Spike account

spike login
spike login status

Sign-in connects the CLI to your Spike account, durable remote Agent features and account-connected Runtimes. It also keeps authenticated requests out of the small anonymous quota.

Keep Spike current

spike update
spike --version

Spike Runtime hosting requires CLI 2.0.17 or newer.

Use a local model directly

Direct OSS mode keeps the CLI on the same machine as the model server and does not use the account Runtime relay.

spike --oss --local-provider lmstudio

spike --oss --local-provider ollama

For another OpenAI-compatible URL, set CODEX_OSS_BASE_URL before starting Spike.

Host that model as Spike Runtime

Runtime makes the provider discoverable to your other signed-in Spike clients without exposing it publicly.

spike_runtime_provider = "lmstudio"

spike runtime serve --label "Home Runtime"

Keep the foreground process running. See the full Spike Runtime guide for Ollama, custom providers and connecting clients.

Troubleshooting

Run the built-in checks

spike doctor
spike login status

Command not found after install

Open a fresh terminal. On macOS and Linux, ensure $HOME/.local/bin is in your PATH.

429 from the chat endpoint

Sign in again if you intended to use the account path. If you intended to use your own model, confirm OSS mode or the connected Runtime is actually selected.

Runtime command is missing

Run spike update and verify the version is 2.0.17 or newer.