OmniRoute MCP Server Documentation
v3.8.1Last updated: 2026-05-13
Was this page helpful?
Loading OmniRoute...
Last updated: 2026-05-13
Was this page helpful?
(30 tools) +
(3 tools) +(4 tools). Tool registration and scope wiring lives in.
diagrams/mcp-tools-37.mmd
omniroute --mcp
# HTTP streamable transport (port 20130) omniroute --dev # MCP auto-starts on /mcp endpoint
factory:
| Transport | Where | When to use |
|---|---|---|
|
|
IDE integrations (Claude Desktop, Cursor, etc.) |
|
via |
Browser/agent clients that need an event stream |
|
|
Multi-session HTTP clients ( header) |
or ) is selected by the setting. Switching transports closes existing sessions on the other transport.
MCP Client Configuration for Claude Desktop, Cursor, Cline, and compatible MCP client setup.
| Tool | Scopes | Description |
|---|---|---|
|
|
Uptime, memory, circuit breakers, rate limits, cache stats |
|
|
All configured combos with strategies (optional metrics) |
|
|
Performance metrics for a specific combo |
|
|
Activate or deactivate a combo |
|
|
Quota used/total, percent remaining, reset time, token health |
|
|
Send a chat completion through OmniRoute routing |
|
|
Cost report by period (session/day/week/month) |
|
|
Full model catalog with capabilities, status, pricing |
| Tool | Scopes | Description |
|---|---|---|
|
|
Web search through OmniRoute search gateway (Serper/Brave/Perplexity/Exa/Tavily/Google PSE/Linkup/SearchAPI/SearXNG) with failover |
| Tool | Scopes | Description |
|---|---|---|
|
, |
Dry-run routing simulation with fallback tree |
|
|
Session budget with degrade/block/alert action |
|
|
Update combo strategy at runtime (priority/weighted/auto/etc.) |
|
|
Apply / |
|
, |
Live test of every provider in a combo using a real upstream call |
|
|
Per-provider metrics with p50/p95/p99 latency and circuit breaker state |
|
, |
Recommend combo by task type with budget/latency constraints |
|
, |
Explain why a request was routed to a provider (scoring factors + fallbacks) |
|
|
Full session snapshot: cost, tokens, top models/providers, errors, budget guard |
|
, |
Diagnose (and optionally auto-repair) database drift like broken combo refs / orphan rows |
|
|
Sync pricing data from external sources (LiteLLM); supports |
| Tool | Scopes | Description |
|---|---|---|
|
|
Semantic cache, prompt-cache, and idempotency stats |
|
|
Flush cache globally or by signature/model |
| Tool | Scopes | Description |
|---|---|---|
|
|
Compression settings, analytics summary, and cache-aware stats (includes metadata) |
|
|
Configure compression mode, threshold, target ratio, system-prompt preservation, MCP description compression toggle |
|
|
Pick the active engine (off/caveman/rtk/stacked) and Caveman/RTK intensity |
|
|
List named compression combos and their engine pipelines |
|
|
Analytics grouped by compression combo and engine |
reports MCP description compression separately under
. Those values are metadata-size estimates for MCP listable
descriptions (, , , and ); they are not provider usage
receipts and are marked with .
tool results of MCP browser/accessibility tools before they are returned to the agent. This filter is not itself a tool — it runs transparently on any tool result that contains verbose accessibility-tree or browser-snapshot text (≥2000 chars).
in global settings (migration 056).
Implementation: .
Full docs: Compression Engines — MCP Accessibility Tree Filter.
Compression Engines and RTK Compression for the runtime compression model behind these tools.
| Tool | Scopes | Description |
|---|---|---|
|
|
Fetch free proxies from the 1proxy marketplace (protocol/country/quality/limit filters) |
|
|
Get the next available proxy by strategy ( / |
|
|
Pool stats, sync status, distribution by protocol and country |
. Auth/scope is enforced through the standard MCP scope pipeline.
| Tool | Description |
|---|---|
|
Search memories by query / type / API key with token-budget enforcement |
|
Add a new memory entry ( / |
|
Clear memories for an API key, optionally filtered by type or timestamp |
. Backed by + .
| Tool | Description |
|---|---|
|
List registered skills with optional filtering by API key, name, or enabled state |
|
Enable or disable a specific skill by ID |
|
Execute a skill with provided input and return the execution record |
|
List recent skill execution history |
) and are not part of the MCP tool catalog — calling a Cloud Agent does not consume an MCP scope.
, , , ).
- ,
, , , .
- docs/frameworks/CLOUD_AGENT.md.
entries) and the guardrails audit trail — a request may be rejected by a guardrail before it ever reaches the MCP scope enforcement layer.
| Endpoint | Method | Description | Auth |
|---|---|---|---|
|
|
Server status: heartbeat, HTTP transport state, audit activity summary | Management (session/admin) |
|
|
Tool catalog (name, description, scopes, phase, source endpoints) | Management |
|
/ |
SSE transport endpoint (gated by + |
API key + scopes |
|
/ |
Streamable HTTP transport (uses header; |
API key + scopes |
|
|
Audit log entries from (filters: |
Management |
|
|
Aggregated audit stats (, |
Management |
.
) and the appropriate is selected. If the wrong transport is configured the route returns HTTP 400 with a hint to switch settings.
. Each tool requires specific scopes:
| Scope | Tools |
|---|---|
|
, |
|
, |
|
, |
|
|
|
, |
|
|
|
, |
|
|
|
|
|
, |
|
|
|
|
|
|
|
, |
|
, |
|
, |
grants all read-scopes, grants full access.
| Variable | Default | Purpose |
|---|---|---|
|
|
Base URL the MCP server uses when calling OmniRoute internal APIs |
|
(empty) | API key forwarded as to internal API calls |
|
(only |
When enabled, missing scopes deny tool calls and log in audit log |
|
(empty) | Comma-separated allowlist of scopes considered "available" by default (used when caller does not provide its own scopes) |
|
(unset = on) | When set to , disables MCP description compression at registration time |
|
(unset = on) | Alternate alias for the same toggle as above |
|
|
Heartbeat file is written to |
and is wired into the MCP server via inside .
settings table (default: enabled) — exposed in the UI as Analytics → MCP description compression.. and tagged to disambiguate from real provider usage receipts. every 5 seconds. The dashboard () reads this file plus PID liveness to derive . HTTP transports report state from in-process instead (no file write).
{
"pid": 12345,
"startedAt": "2026-05-13T12:34:56.000Z",
"lastHeartbeatAt": "2026-05-13T12:35:01.000Z",
"version": "1.8.1",
"transport": "stdio",
"scopesEnforced": false,
"allowedScopes": [],
"toolCount": 37
}
table by :
and REST endpoints to inspect recent calls.
| File | Purpose |
|---|---|
|
MCP server factory, stdio entry point, scoped tool registrations |
|
SSE + Streamable HTTP transport (session management) |
|
Tool scope evaluation and caller resolution |
|
Tool call audit logging () |
|
stdio heartbeat writer () |
|
Description compression for tool / prompt / resource registries |
|
Zod schemas + tool registry (, 30 entries) |
|
Phase 2 + cache + 1proxy tool handlers |
|
Compression tool handlers |
|
Memory tool definitions (3 tools) |
|
Skill tool definitions (4 tools) |
|
endpoint |
|
endpoint |
|
SSE transport route |
|
Streamable HTTP transport route |
|
audit log query |
|
aggregated audit metrics |