reliapi/config.yaml at main · kiku-jw/reliapi · 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
# ReliAPI Configuration Example
# All targets (upstreams) are defined here

targets:
  # Example: OpenAI LLM provider
  openai:
    base_url: "https://api.openai.com/v1"
    timeout_ms: 20000
    circuit:
      error_threshold: 5
      cooldown_s: 60
    llm:
      provider: "openai"  # Explicit provider (optional, auto-detected from base_url if not specified)
      default_model: "gpt-4o-mini"
      max_tokens: 1024
      temperature: 0.7
      # Budget control: predictable costs
      soft_cost_cap_usd: 0.01    # Warn and throttle if exceeded
      hard_cost_cap_usd: 0.05    # Reject if exceeded
    cache:
      ttl_s: 60
      enabled: true
    auth:
      type: bearer_env
      env_var: OPENAI_API_KEY
    # fallback_targets: ["anthropic", "mistral"]  # Planned: simple fallback chain for LLM proxy
    retry_matrix:
      "429":
        attempts: 3
        backoff: "exp-jitter"
        base_s: 1.0
        max_s: 60.0
      "5xx":
        attempts: 2
        backoff: "exp-jitter"
        base_s: 1.0
      "net":
        attempts: 2
        backoff: "exp-jitter"
        base_s: 1.0

  # Example: Anthropic LLM provider
  anthropic:
    base_url: "https://api.anthropic.com/v1"
    timeout_ms: 25000
    circuit:
      error_threshold: 5
      cooldown_s: 60
    llm:
      provider: "anthropic"  # Explicit provider (optional, auto-detected from base_url if not specified)
      default_model: "claude-3-haiku-20240307"
      max_tokens: 2048
      temperature: 0.2
      # Budget control
      soft_cost_cap_usd: 0.005
      hard_cost_cap_usd: 0.02
    cache:
      ttl_s: 300
      enabled: true
    auth:
      type: bearer_env
      env_var: ANTHROPIC_API_KEY
    retry_matrix:
      "429":
        attempts: 3
        backoff: "exp-jitter"
        base_s: 2.0
      "5xx":
        attempts: 2
        backoff: "exp-jitter"
        base_s: 1.0

  # Example: Generic HTTP API (not LLM)
  payments_api:
    base_url: "https://api.payments.example.com"
    timeout_ms: 12000
    circuit:
      error_threshold: 3
      cooldown_s: 30
    cache:
      ttl_s: 300
      enabled: true
    auth:
      type: api_key
      header: "X-API-Key"
      env_var: PAYMENTS_API_KEY
    retry_matrix:
      "429":
        attempts: 3
        backoff: "exp-jitter"
        base_s: 1.0
      "5xx":
        attempts: 2
        backoff: "exp-jitter"
        base_s: 1.0