Core and optional components
Fased keeps the Agent and crypto path in core. Large chat integrations and machine-specific runtimes are installed only when you choose them. Inspect the current machine from the CLI:included, installed, configured, and ready are separate states. A
configuration field does not mean its runtime is installed or healthy.
Each component row exposes only applicable actions:
- Install downloads an official npm add-on.
- Restart reloads an installed component through the Gateway service.
- Connect opens the owning setup surface.
- Open docs opens the canonical setup page.
Included in core
A normal install includes:- Gateway, Control UI, browser Chat, Agents, Tasks, Tools, and provider clients
- Solana wallet controls and the singleton Mining wallet flow
- the bundled
sat-miningruntime, SAT shared specs, Mining UI, and mining CLI - Fased Network, offers, requests, trust, bond, and marketplace controls
- browser-control interfaces and the Fased browser extension files
- file-backed memory and remote embedding-provider support
sat-mining runtime is included and loaded with core. Creating or
importing @wallet:mining through onboarding attaches the dedicated Mining
wallet and makes readiness checks available. Do not install a separate mining
package and do not run fased plugins enable sat-mining as the normal mining
setup path. Mining does not start until readiness, capital, and commit checks
pass and the operator explicitly starts it.
Official channel add-ons
Telegram, WhatsApp, Discord, Slack, Feishu/Lark, and Google Chat are official npm add-ons. They are not downloaded by a fresh core install. The normal browser flow is:- Open Agents > selected Agent > Channels.
- Select the channel.
- Click Install.
- Restart the Gateway when prompted.
- Return to the channel card and enter its token or complete its login flow.
fased channels add offer the same npm download when
you select one of these channels.
CLI installation is also available:
The packages above are the official installable channel add-ons in this
release. Other advanced channel extensions may need a source install or their
own channel-specific dependencies until they receive a published add-on
package. Follow the individual channel page instead of assuming the lightweight
hosted runtime downloads every advanced channel dependency.
Optional runtime add-ons
Install these only when the related feature is needed:- Browser Runtime adds Playwright control and readable-page extraction.
- Media Runtime adds image transforms, file-type detection, and PDF extraction.
- Speech Runtime adds the Edge TTS client. OpenAI and ElevenLabs API speech do not require this local package.
- Local Vector Memory adds native sqlite vector acceleration. File-backed memory and remote embedding providers remain in core.
- OpenAI Sign-In Runtime adds the official OpenAI Codex app-server used to discover and execute models available to a ChatGPT sign-in. Direct OpenAI API key models do not require this component.
fased update can update them after the core update succeeds.
Local models stay external
Fased includes provider clients for Ollama, LM Studio, vLLM, LiteLLM, and custom OpenAI-compatible endpoints. It does not install those model servers and does not download model weights.- Install and run the model server on the machine that owns the model.
- Download or load the model with that server.
- Connect it from Agent > Models.
Browser control has two optional layers
Fased includes the browser-control interfaces and extension files. Install the Browser Runtime component for Playwright control and readable-page extraction:- an existing system Chromium browser
- the Fased extension relay
- a separately installed Chromium browser for the managed
fasedprofile - remote CDP or a paired node host when the Gateway runs on a VPS
Local memory embeddings are optional
Native sqlite vector acceleration is a separate component:memorySearch.provider = "local" uses the native node-llama-cpp runtime and a
GGUF embedding model. The lightweight hosted runtime does not include that
native add-on or its model file.
Use a source install when you intentionally maintain the native local-embedding
stack. Otherwise choose Ollama or a remote embedding provider. This boundary is
separate from using Ollama or LM Studio as the Agent’s chat model.
See Memory.
Install and update behavior
After the core update succeeds,
fased update checks tracked npm plugins and
updates compatible add-ons. You can review them manually with: