Straits Industriesproduction
Search models, nodes, traces…⌘K
SI

gen_sf42af43ym

succeededbypass

chat request through @company/auto — the complete decision record: guardrails, routing, fallback, cache and settlement.

Principal
Ops Copilot Agent(agent)
Project
Agent Operations
Profile
@company/auto
Requested
fabric/auto
Resolved model
anthropic/claude-sonnet
Session
sess_agent_ops_01
Tokens in / out
12K / 1.4K
Cached tokens
10K
Cost
$0.03
Latency
1.68 s
Attempts
1
Age
2d ago

Decision timeline

  1. admission+0 ms

    chat request by prin_agent under profile @company/auto (requested "fabric/auto")

  2. PRE_ROUTE+7 ms
    allowprompt-injection-heuristictool-loop prompt, session pinned
  3. cache+7 ms

    cache bypassed (policy off or non-cacheable request)

  4. route+7 ms

    selected anthropic/claude-sonnet via anthropic/acct_anthropic_ent_apse1 on node_sg_dc_01 (C-tier, ap-southeast-1) — objective=balanced, 3 candidates scored, 2 in fallback chain, 1 hard-filter rejection [router router-v3.4.1, policy policy-2026-07-14]

    4 candidates considered3 survived hard filters1 rejectedobjective: balancedfallback chain: 2router-v3.4.1 · policy-2026-07-14
    Hard-filter rejections
    • ep_gpt5nano_openai_use1@node_cf_use1_01context_too_long: 142000 tokens exceeds model context 128000
    Score breakdown — top 3
    #CandidatequalitycostlatencyPenaltycacheSavingscontractValueshadowPriceretryRiskaffinityPenaltyTotal
    1anthropic/claude-sonnet via anthropic · node_sg_dc_01 selected+0.561-0.270-0.004+0.033+0.021-0.002-0.003+0.3000.636
    2openai/gpt-5.2 via openai · node_sg_dc_01 +0.575-0.169-0.024+0.0350.000-0.010-0.003-0.1500.240
    3anthropic/claude-sonnet via amazon-bedrock · node_sg_dc_01 +0.567-0.297-0.044+0.042+0.020-0.002-0.003-0.1500.104
  5. attempt #1+35 ms

    #1 anthropic · account acct_anthropic_ent_apse1 · node node_sg_dc_01 — succeeded in 1591ms, 10193 prompt-cached tokens

    succeededanthropicacct_anthropic_ent_apse1pool enterprisenode_sg_dc_01Tier C1.59 s

    12K in / 1.4K out / 10K cached

  6. settlement+1682 ms

    settled $0.03 — 11667 in / 1399 out / 10193 cached tokens (13066 total)

    Ledger entries
    • reserve$0.05budget:bud_proj_agentsreserve est max for fabric/auto
    • settle$0.03budget:bud_proj_agentssettle actual provider usage
    • refund−$0.02budget:bud_proj_agentsrefund unused reservation
    Cost layers (memo §15.3)
    Provider list
    $0.03
    Contract-adjusted
    $0.03
    Internal charge
    $0.03
    Contract savings
    $0.0023

    contract ct_anthropic_committed (committed)

  7. complete+1682 ms

    status=succeeded, model=anthropic/claude-sonnet, 1682ms end-to-end

fabric_metadata

The route-explanation envelope an API caller receives with X-OpenRouter-Metadata: enabled or routing.explain, reconstructed from this generation record.

fabric_metadata.json
{
  "requested": "fabric/auto",
  "resolved_model": "anthropic/claude-sonnet",
  "strategy": "balanced",
  "policy_version": "policy-2026-07-14",
  "router_version": "router-v3.4.1",
  "provider": "anthropic",
  "region": "ap-southeast-1",
  "account_pool": "enterprise",
  "node": {
    "ref": "node_sg_dc_01",
    "trust_tier": "C"
  },
  "cache": {
    "response": "BYPASS",
    "prompt_affinity": "HIT",
    "predicted_cached_tokens": 10193
  },
  "guardrails": [
    {
      "stage": "PRE_ROUTE",
      "action": "allow",
      "profile": "gp_baseline",
      "evaluations": 1
    }
  ],
  "attempts": [
    {
      "provider": "anthropic",
      "status": 200
    }
  ],
  "routing_reason": [
    "selected anthropic/claude-sonnet via anthropic (balanced objective, score 0.636)",
    "committed capacity preferred (contract economics)",
    "1 candidate(s) rejected by hard filters"
  ]
}