Tera Router
Getting Started

Dashboard Tour

Get to know every page in the Tera Router dashboard — from API Keys to Console Log.

The dashboard (apps/web-ui) is the only place you need for day-to-day router management. Every page below calls the same server REST API — if you prefer automation, everything can be called directly (see the API Reference).

Overview

  • Dashboard — traffic, spend, and routing performance at a glance.

Traffic

  • Endpoints — how to connect your apps: base URL + API key, plus one-click tunnels for access from other machines: a public Cloudflare quick tunnel or a private Tailscale address (see Remote access).
  • Chains — build model chains: sequential steps that can each be switched on or off, strategies (priority, round-robin, load-balanced), and fallbacks.
  • Skills — reusable prompt snippets, plus reference skills you can copy to a coding agent so it knows how to use the router. Any skill opens in a rendered preview.
  • Model Alias — one alias name mapped to an ordered pool of provider/model targets. Keys and plans reference the alias, so swapping providers never needs re-assignment.
  • Pricing — per-model token rates the gateway charges with. Overrides beat the built-in catalog; a provider-wide override prices every model without its own row.

Connections

  • API Keys — create, restrict models, disable, and reveal sk_tr_ keys. Each key can be bound to a plan.
  • Providers — connect provider accounts (API key, or official sign-in for OpenAI and Anthropic), set priorities, test connections, and view per-account quotas. Tabs split the catalog into AI Model, AI Search, AI Voice, and AI Image Generation providers.
  • Media — the same catalog grouped by non-chat capability: embeddings, image, TTS, STT, web search, and web fetch.
  • Proxy Pools — group outbound proxies for accounts that need specific IPs.

Safety

  • Guardrails — PII, prompt injection, topic, toxicity, and bias policies, layered from global down to per-key. Includes a dry-run mode.
  • Provider Health — the condition of each provider account: cooldowns, failures, and quota status.

Cost & Analytics

  • Usage — requests, tokens, cost, and latency per time range (today, 24 hours, 7 days, 30 days), broken down by model and client (Claude Code, Cursor, Codex, etc).
  • Plans — limit templates: spend, tokens, period, alerts, hard cutoff, and model allowlists.
  • Budgets — spend and token caps per period for the whole instance, one API key, or one provider account; alert on or block traffic once a cap is hit.
  • Quota Tracker — usage against active budgets, pausable per item.
  • System — host and server process metrics.
  • Settings — router behavior: timeouts, round-robin provider/chain, outbound proxy, request detail recording, token saving (tool-output compression, an optional Headroom proxy, and terse-output directives), down to display branding.

Developer

  • Playground — chat with any provider model, model alias, or chain through the gateway, with streaming. Each reply shows which provider and model actually answered.
  • Console Log — a live stream of recent server logs (500-entry ring buffer).
  • CLI Tools — one-click configuration for coding tools (Claude Code, Cursor, and others) pointing straight at your Tera Router instance. The Claude Code page detects the claude CLI on the server host, lets you pick an endpoint, API key, and optional default model (provider/model-id or chain:<name>), and either writes them into ~/.claude/settings.json on that host (Apply, merging with existing settings; honors CLAUDE_CONFIG_DIR) or copies a ready-to-paste snippet for another machine.

One page, one concept

If this is your first time: start with Providers (connect accounts), then API Keys (create access), then Endpoints (connect your apps). The rest follows as needed.

On this page