mcprouter/config.example.yaml at master · JSLEEKR/mcprouter · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
# mcprouter config — annotated example
#
# This file shows every option mcprouter supports. Remove sections you
# do not need. Durations accept Go-style strings (e.g. "5s", "2m").

server:
  host: 127.0.0.1
  port: 8080
  read_timeout: 30s
  write_timeout: 60s
  idle_timeout: 120s
  shutdown_timeout: 10s

# Backends are the upstream MCP servers mcprouter will fan requests out to.
backends:
  # Local filesystem MCP server (stdio transport)
  - name: fs
    transport: stdio
    command: npx
    args: ["-y", "@modelcontextprotocol/server-filesystem", "/workspace"]
    priority: 10
    weight: 2
    # Capability hints. Optional — discovery will fill these in too.
    tools: [read_file, write_file, list_directory]
    resources: ["file://"]
    health:
      interval: 30s
      timeout: 5s
      method: ping
    timeout: 30s

  # GitHub MCP server (remote HTTP transport with bearer auth)
  - name: github
    transport: http
    url: https://mcp.example.com/github/rpc
    priority: 8
    weight: 1
    tools: [create_issue, get_pr, list_repos]
    auth:
      type: bearer
      token: ghp_YOUR_TOKEN_HERE
    health:
      interval: 60s
      timeout: 10s

  # Postgres MCP server (SSE transport with API key auth)
  - name: postgres
    transport: sse
    url: https://mcp.example.com/postgres/sse
    tools: [query, describe_table, list_databases]
    auth:
      type: api_key
      header: X-API-Key
      token: pg_SECRET_KEY
    health:
      interval: 30s
      timeout: 5s

  # Backup filesystem used only if 'fs' goes down (see fallback below)
  - name: fs-backup
    transport: stdio
    command: node
    args: ["./backup-fs-server.js"]
    weight: 1
    tools: [read_file, write_file]
    health:
      interval: 30s
      timeout: 5s

routing:
  # Strategy used when multiple backends can serve the same capability.
  # Options: round_robin | weighted | least_connections | sticky
  strategy: round_robin

  # When true, aggregated tools/list responses are namespaced with
  # "<backend><sep><tool>" so clients can disambiguate duplicates.
  namespace: true
  namespace_sep: "__"

  # If no capability-based routing matches, send to this backend.
  default_backend: fs

# Fallback chains: ordered lists of backends to try for a capability.
# Capability keys:
#   "tool:NAME"  — fallback for a specific tool name
#   "tools/call" — fallback for every tools/call request
fallback:
  - capability: "tool:read_file"
    chain: [fs, fs-backup]
    retries: 2
    backoff: 200ms
  - capability: "tool:create_issue"
    chain: [github]
    retries: 1
    backoff: 500ms

logging:
  file: /var/log/mcprouter/requests.log
  level: info
  max_size_mb: 50
  max_backups: 5
  redact: true   # scrub token/password/api_key fields from logged params

metrics:
  enabled: true
  path: /metrics