Skip to main content
Glama

TensorFeed

by ai.tensorfeed
OverviewDiscussionsAdmin

route_verdict

TensorFeed's signed model-routing decision: the single best model for a task or named model, fused from live pricing, contamination-discounted benchmarks, real production usage, measured p95 latency, incident state, and deprecation flags, with the reasoning. tier='preview' is free (10 calls per day per IP), top verdict only. tier='full' costs 1 credit ($0.02) and adds ranked runners-up, constraint filters, and an AFTA-signed receipt you can audit. Omit tier and you get the free preview, unless the call carries a payment or credits token (or uses the strict x402 transport), in which case it bills as full. Pay per call with an x402 wallet payment (arguments.payment, or an X-PAYMENT header; strict HTTP-402 transport at https://mcp.tensorfeed.ai/mcp?x402=strict) or use an Authorization: Bearer tf_live_... credits token. No token and no USDC? Claim free trial credits by signing a wallet message at https://tensorfeed.ai/api/payment/trial-credits.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
taskNoTask type to route for (code, reasoning, creative, general). Provide task or model.
tierNo'preview' (free, 10 calls per IP per day) or 'full' (1 credit; adds ranked runners-up, constraint filters, signed receipt). Omitted: preview, unless a payment or credits token is attached.
modelNoModel id or display name to narrow the verdict to one model (e.g. "Claude Opus 5.5" or "claude-opus-5-5"). Provide task or model.
budgetNoFull tier only. Max blended USD per 1M tokens.
paymentNoOptional base64 x402 payment payload (the same string an X-PAYMENT header would carry). Sign against the accepts[0] requirement returned by an unpaid call. Alternative to the Bearer credits token.
min_qualityNoFull tier only. Minimum quality in [0, 1] as a share of the best candidate measured on the same benchmark set (0.9 keeps models within 10 percent of the top).
exclude_deprecatedNoFull tier only. Default true. Pass "false" to keep deprecated or sunsetted models.
max_latency_p95_msNoFull tier only. Drop candidates whose measured p95 latency exceeds this value (ms).
require_operationalNoFull tier only. Default true. Pass "false" to keep candidates known down or in failover.

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • changedInput schema / properties / min_quality / description
      Previous value: -"Full tier only. Minimum trust-discounted quality score in [0, 1]."New value: +"Full tier only. Minimum quality in [0, 1] as a share of the best candidate measured on the same benchmark set (0.9 keeps models within 10 percent of the top)."
  2. Changed9 schema fields changed
    • addedInput schema / properties / budget
      Added value: +{
+  "description": "Full tier only. Max blended USD per 1M tokens.",
+  "type": "number"
+}
    • changedInput schema / properties / exclude_deprecated / description
      Previous value: -"Default true. Pass \"false\" to keep deprecated or sunsetted models."New value: +"Full tier only. Default true. Pass \"false\" to keep deprecated or sunsetted models."
    • changedInput schema / properties / max_latency_p95_ms / description
      Previous value: -"Drop candidates whose measured p95 latency exceeds this floor."New value: +"Full tier only. Drop candidates whose measured p95 latency exceeds this value (ms)."
    • addedInput schema / properties / min_quality
      Added value: +{
+  "description": "Full tier only. Minimum trust-discounted quality score in [0, 1].",
+  "type": "number"
+}
    • changedInput schema / properties / model / description
      Previous value: -"Canonical model id or display name to narrow the verdict to one model."New value: +"Model id or display name to narrow the verdict to one model (e.g. \"Claude Opus 5.5\" or \"claude-opus-5-5\"). Provide task or model."
    • changedInput schema / properties / require_operational / description
      Previous value: -"Default true. Pass \"false\" to keep candidates known down or in failover."New value: +"Full tier only. Default true. Pass \"false\" to keep candidates known down or in failover."
    • changedInput schema / properties / task / description
      Previous value: -"One of code, reasoning, creative, general. Provide task or model."New value: +"Task type to route for (code, reasoning, creative, general). Provide task or model."
    • addedInput schema / properties / task / enum
      Added value: +[
+  "code",
+  "reasoning",
+  "creative",
+  "general"
+]
    • addedInput schema / properties / tier
      Added value: +{
+  "description": "'preview' (free, 10 calls per IP per day) or 'full' (1 credit; adds ranked runners-up, constraint filters, signed receipt). Omitted: preview, unless a payment or credits token is attached.",
+  "enum": [
+    "preview",
+    "full"
+  ],
+  "type": "string"
+}
  3. Changed1 schema field changed
    • addedInput schema / properties / payment
      Added value: +{
+  "description": "Optional base64 x402 payment payload (the same string an X-PAYMENT header would carry). Sign against the accepts[0] requirement returned by an unpaid call. Alternative to the Bearer credits token.",
+  "type": "string"
+}
  4. Added

TDQS

A4.5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description carries the full burden and does so well: it discloses cost (1 credit / $0.02), rate limits (10 calls/day/IP on preview), billing triggers when a payment or credits token is attached, auth mechanisms (x402 payment, X-PAYMENT header, Bearer tf_live_ token, strict transport URL), and a free-trial path. These are exactly the pre-invocation risks an agent must know.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose is front-loaded and the billing/payment detail, though dense, is invocation-relevant rather than filler. There is some redundancy (preview's freeness and the omitted-tier default are stated twice), and the run-on payment sentence is hard to parse, keeping it below a 5.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With 9 parameters, no annotations, and no output schema, the description does the heavy lifting on billing, auth, and tier behavior, and sketches the return (top verdict with reasoning, ranked runners-up, signed receipt). It stops short of fully describing the verdict payload, but covers everything needed to invoke it without surprise charges.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so the baseline is 3, but the description adds real value beyond the schema: the cross-parameter default logic (omitting tier yields preview unless a payment/credits token is present, which bills as full) is not derivable from the individual field descriptions. It does not, however, restate or clarify the full-tier-only constraint fields beyond what the schema already says.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence gives a precise verb+resource: a signed model-routing decision returning the single best model for a task or named model, fused from named signals. The scope ('single best') is inherently distinct from siblings like compare_models or failover_verdict, so an agent can route correctly without opening the schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Strong guidance on tier selection: preview vs full, what each returns, and the default behavior when tier is omitted. However, it never states when to prefer this tool over sibling verdict tools (compare_models, provider_reliability_verdict, failover_verdict), so the cross-tool selection is left to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All TensorFeed toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources