Switchboard
by gelevanog
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| DATABASE_URL | No | owner connection | local PostgreSQL |
| SWITCHBOARD_CIMD | No | Client ID Metadata Documents | true |
| SWITCHBOARD_TODAY | No | the demo's 'today' (empty = the real date) | 2026-10-01 |
| SWITCHBOARD_TOKEN | No | a token for the user when running the gateway in stdio mode (from `switchboard token --user …`) | |
| ANALYTICS_MAX_ROWS | No | row cap of analytics queries | 200 |
| OPENROUTER_API_KEY | No | API key for OpenRouter (used by the playground and the evaluation) | none |
| SWITCHBOARD_LLM_MODEL | No | model for the playground and the evaluation | nvidia/nemotron-3-super-120b-a12b:free |
| ANALYTICS_DATABASE_URL | No | the read-only login for analytics queries | derived |
| SWITCHBOARD_DEMO_LOGIN | No | the demo sign-in page (pick a user, no password); set false once a real identity provider is wired in | true |
| SWITCHBOARD_LLM_LEDGER | No | ledger of real API calls | results/calls.jsonl |
| SWITCHBOARD_PUBLIC_URL | No | the URL clients use; issuer of every token; the MCP resource is <url>/mcp | http://localhost:8080 |
| SWITCHBOARD_POLICY_FILE | No | tenants, upstream servers, roles (tool patterns and scopes), confirmation rules, rate limits, untrusted-content handling | config/switchboard.yaml |
| SWITCHBOARD_LLM_MAX_CALLS | No | hard budget | 500 |
| SWITCHBOARD_LIST_PAGE_SIZE | No | page size of tools/list and resources/list | 20 |
| SWITCHBOARD_ALLOWED_ORIGINS | No | browser origins allowed to call /mcp | console, Inspector |
| SWITCHBOARD_ACCESS_TOKEN_TTL | No | access-token lifetime in seconds (refresh tokens: 30 days, rotated) | 900 |
| ANALYTICS_STATEMENT_TIMEOUT_MS | No | time limit of analytics queries | 5000 |
| SWITCHBOARD_REQUIRE_FREE_MODELS | No | refuse OpenRouter model ids without :free, and answers served by a paid model | true |
| SWITCHBOARD_ALLOW_HTTPS_REDIRECTS | No | dynamic registration with https redirect URIs (claude.ai, ChatGPT) | true |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Server capabilities have not been inspected yet.
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
No tools | |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
This server cannot be deployed
Maintenance
ActivityMaintained
ResponsivenessNo issues