> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fased.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Control UI Setup Model
# Control UI Setup Model
Onboarding gets the machine ready. The Control UI makes the agent useful.
The intended flow is:
```mermaid theme={"theme":{"light":"min-light","dark":"min-dark"}}
flowchart TD
onboard["Onboarding"] --> models["Agent > Models"]
models --> skills["Agent > Skills"]
skills --> channels["Agent > Channels"]
channels --> services["Agent > Services"]
services --> memory["Agent > Memory"]
memory --> tasks["Agent > Tasks"]
tasks --> operate["Daily operation"]
classDef machine fill:#120605,stroke:#ff5a36,color:#ffffff;
classDef agent fill:#071018,stroke:#12cfff,color:#ffffff;
classDef run fill:#20120a,stroke:#ffb020,color:#ffffff;
class onboard machine;
class models,skills,channels,services,memory,tasks agent;
class operate run;
```
The first browser setup pass should start from the selected Agent:
Create or select the Agent that will own chat, chat app routes, tasks, and saved context.
Connect provider auth if needed, then choose primary, fallback, and task models.
Create, review, install, configure, and allow skills for this Agent.
Install the official channel add-on when required, then connect and route
accounts, topics, or guilds to this Agent.
Connect external APIs such as web/search, Gmail, Calendar, GitHub, or browser/media.
Enable saved session context and review per-Agent archive state.
Create saved tasks, triggers, workflows, graphs, programs, and templates.
Use Advanced Config only for fields that do not have a friendly page yet.
For the focused model-to-agent flow, see
[Models to Agents to Chat](/start/provider-agent-chat-flow).
## What Onboarding Owns
Use onboarding for machine and security setup:
* Local or Hosting profile
* Gateway mode, port, auth, and token
* workspace and service startup
* local signer wallet setup when selected
* optional singleton Mining wallet setup
* Hosting profile Tailscale and host hardening
Onboarding is not the normal place to manage every model provider, skill,
extension, hook, channel, service, or scheduled task.
Hosting is for VPS or always-on server machines. It can change SSH and firewall
behavior. Local is for a laptop, desktop, or dev box and does not apply the
hosting security baseline.
## What The Control UI Owns
After onboarding, open:
```bash theme={"theme":{"light":"min-light","dark":"min-dark"}}
fased dashboard
```
Then use the browser pages for product setup:
Create or select an Agent and attach model, skills, services, chat apps,
saved context, tasks, tools, sessions, and wallet controls.
Review runtime plugins, source trust, dependencies, scanner warnings, and advanced extension setup.
Review local model usage by provider, model, Agent, session, task, chat app, and source.
Configure notification routing and recent in-app/external delivery.
Review wallet controls, approvals, balances, custody, and signer health.
Operate SAT mining readiness, capital, commit, recovery, and history.
Operate Fased Network status, routing, and marketplace-facing identity.
Review Fased Network offers, requests, purchases, and disputes.
Read memory diagnostics across Agents. Enable session-memory from `Agent > Memory`.
Advanced Config only. Use this when a field is not exposed in a friendly page yet.
Dashboard is a launch/status widget board. It is not the setup wizard. Start
with `/agents` when the Gateway is already online.
## Agent Means
In the Control UI, an Agent is the workspace profile you actually operate.
It can have:
* model and fallbacks
* skills
* services
* chat apps
* saved context settings
* scheduled tasks
* files and workspace context
* wallet controls
Examples:
* Researcher
* Support
* Mining Operator
* Operations Reviewer
Channels and scheduled tasks should route to a configured Agent. In the UI,
Channels means chat app connections such as Telegram, Discord, and WhatsApp.
Subagents are different: they are internal runtime workers used during a task,
not normal chat routing targets.
`Agent > Tasks` is the selected Agent's saved-work control surface. Create a
**Task** for scheduled work first. Add Triggers, Workflows, Graphs, Programs,
or Templates only when the run needs that structure.
Run history can include scheduled runs plus records mirrored from webhooks,
chat apps, subagents/ACP, CLI/system runs, media generation, wallet approvals,
Marketplace, mining, and workflows. It is audit data, not the primary task list.
Use the owning page for the actual domain operation: Wallets for signing,
Marketplace for orders/disputes, Mining for mining control, Channels for chat
routes/delivery, and Services/Media for connector setup.
## Skills, Extensions, Channels, Services
Use these words consistently:
**Skill**
A user-facing ability or instruction pack the agent can use. Create, review,
configure, and allow it from `Agent > Skills`.
**Extension**
Plugin code that can add tools, channels, hooks, schemas, commands, or UI
panels. Review it in `/extensions`.
**Channel**
A chat app where the agent receives or sends messages. Connect and route it from
`Agent > Channels`. Channels do not own tasks.
**Service**
An external API the agent can use, such as Gmail, Calendar, GitHub, web/search,
browser/media, or a plugin-reported API. Connect it from `Agent > Services`.
**Hook**
Background automation around Fased events. Review hook packs in `/extensions`;
Agent memory archive control lives in `Agent > Memory`.
Skill setup is intentionally split by responsibility but exposed in one Agent
surface:
* **Agent > Skills**: create workspace skills, review plugin-catalog installs, save
skill config, fix dependencies, edit skill files, and allow/deny skills for
that Agent.
* **Agent > Tools**: allow or deny tools exposed by core code, services,
channels, or extensions.
* **Agent > Services**: connect API credentials a skill/tool depends on.
* **Wallet > Skill Grants tab**: grant narrow Agent-wallet actions for reviewed
wallet-capable skills. Mining and Vault wallets stay outside generic skill
grants.
`Agent > Channels` exposes installed channels and official add-ons in the same
order as onboarding. Telegram, WhatsApp, Discord, Slack, Feishu, and Google Chat show an **Install**
action on a fresh core install; setup fields appear after the package is
installed and the Gateway restarts. The normal setup path is per Agent; any
global channel page is an implementation/deep-link fallback, not the first-run
path.
Fased includes provider clients for local model servers, but it does not install
Ollama, LM Studio, vLLM, LiteLLM, or model weights. Install and run that service
first, then connect it from **Agent > Models**. See
[Core And Optional Components](/install/components).
`Agent > Services` is the connector recovery center for task preflight. If a
task needs web search, GitHub, Gmail/Google Workspace, Firecrawl, model auth,
channel delivery, or a configured skill, it should pause in `needs-access` and
point back to the matching setup surface instead of repeatedly running. Access
is still constrained by that Agent's Tools, Skills, wallet controls, and task
policy.
## Memory (Saved Context)
Memory has multiple pieces:
* `MEMORY.md`: curated long-term workspace memory
* `memory/`: markdown archive directory
* `session-memory`: hook that archives recent session messages on `/new` and `/reset`
* QMD/backend: optional indexing, search, and export
* Memory Doctor: diagnostics and repair surfaces
Use `Agent > Memory` to enable or disable `session-memory` and review that
Agent's archive/QMD state. Use `/memory` for cross-Agent diagnostics and memory
health. Repair and raw doctor actions stay in Debug or CLI.
## Advanced Config
`/config` edits `~/.fased/fased.json` directly.
Use it when:
* a plugin or channel exposes a schema field that does not have a friendly UI yet
* you need raw JSON mode
* you need validation, diff, reload, save, or apply controls
* support or docs asks you to change a specific config field
Do not use Advanced Config as the normal first-run setup path. Prefer
Agents, Extensions, Wallets, Mining, Fased Network, Marketplace, Usage,
Notifications, Logs, and Debug first.
## CLI
CLI remains useful for:
* automation
* repair
* export/import
* host-level operations
* dangerous wallet or signer operations
* scripted installs
Normal product setup after onboarding should happen in the Control UI whenever
the relevant page exists.