@tensorfeed/mcp-server 1.36.2 → 1.36.4

package/CHANGELOG.md

@@ -2,6 +2,34 @@
 
 All notable changes to the [TensorFeed.ai MCP server](https://github.com/RipperMercs/tensorfeed-mcp). Free tools work without configuration; premium tools require a bearer token via the `TENSORFEED_TOKEN` env var. Buy credits at [tensorfeed.ai/developers/agent-payments](https://tensorfeed.ai/developers/agent-payments).
 
+## 1.36.4 - 2026-05-31
+
+Docs-only release. The README now opens with the Route Verdict quickstart: a
+zero-install curl against the free preview plus the `route_verdict_preview` and
+`route_verdict` tools. No tool, pricing, or behavior changes from 1.36.3.
+
+## 1.36.3 - 2026-05-31
+
+Adds the Route Verdict wedge: TensorFeed's signed routing decision is now a
+tool you can try free, then upgrade. Brings the catalog to 61 tools (37 free,
+24 premium).
+
+Free tool:
+- `route_verdict_preview`: TensorFeed's signed routing decision that fuses
+  live pricing, contamination-discounted benchmarks, real production usage,
+  measured p95 latency, incident state, and deprecation flags into the single
+  best model for a task or a named model, with the reasoning. No key, 10 calls
+  per day per IP.
+
+Premium tool:
+- `route_verdict` (1 credit): the same decision plus ranked runners-up,
+  constraint filters (max p95 latency, budget, min quality, operational-only,
+  exclude-deprecated), and an AFTA-signed receipt over the exact inputs so you
+  can prove why you routed. Strict premium, no free trial.
+
+The older `premium_routing` (a ranked list with a score breakdown) is
+unchanged and remains available.
+
 ## 1.36.2 - 2026-05-31
 
 Pricing-copy and catalog-accuracy corrections. No change to the free/paid

package/README.md

@@ -6,6 +6,59 @@
 
 This subfolder remains in the main `tensorfeed` repo as the **publishing source** for the npm package (`@tensorfeed/mcp-server`) and the [official MCP registry](https://registry.modelcontextprotocol.io/v0/servers/ai.tensorfeed/mcp-server) entry. Edits to `src/`, `server.json`, `package.json`, etc. happen here and get pushed to the standalone repo on release.
 
+## Featured: Route Verdict
+
+The single best model to use right now, as one signed call. `route_verdict` fuses live pricing, contamination-discounted benchmark capability, real production usage, measured p95 latency probes, live incident state, and deprecation flags into one ranked decision, with an AFTA-signed receipt over the exact inputs. Instead of stitching together pricing pages, benchmark leaderboards, status dashboards, and your own latency tests, you get a current, defensible routing answer in one request.
+
+### Zero install, no key, one command (works today)
+
+```bash
+curl -s -A "tensorfeed-cc-quickstart" "https://tensorfeed.ai/api/preview/route-verdict?task=code"
+```
+
+Swap `task` for `reasoning`, `creative`, or `general`, or pass `?model=<id-or-name>` to score a specific model. The free preview is 10 calls per day per IP, no token. Abridged real response:
+
+```json
+{
+  "ok": true,
+  "preview": true,
+  "query": { "task": "code", "model": null },
+  "verdict": {
+    "rank": 1,
+    "model": { "name": "Gemini 2.5 Pro", "provider": "google" },
+    "pricing": { "blended": 5.625, "unit": "per 1M tokens" },
+    "quality": { "trust_discounted": 0.6498 },
+    "latency": { "measured_p95_ms": 1223, "source": "measured_probe" },
+    "operational": { "ok": true, "status": "operational" },
+    "composite_score": 0.8449,
+    "why": "code quality 0.6498 after trust discount; corroborated by real usage (rank 5, 6.5% share, flat); measured p95 1223 ms; operational; blended $5.625 / 1M"
+  },
+  "rate_limit": { "limit": 10, "remaining": 9, "scope": "per IP per UTC day" },
+  "upgrade": {
+    "premium_endpoint": "/api/premium/route-verdict",
+    "adds": ["runners_up", "AFTA-signed receipt", "filter params", "no rate limit"]
+  }
+}
+```
+
+### The agent path (MCP)
+
+With `@tensorfeed/mcp-server` installed, an agent gets two tools. Call the free one for the pick, then the paid one when it needs to defend the choice:
+
+```text
+# Free taste: the top pick + reasoning, no token (10/IP/day)
+route_verdict_preview({ task: "code" })
+
+# 1 credit: ranked runners-up, constraint filters, AFTA-signed receipt
+route_verdict({ task: "code", max_latency_p95_ms: 1500, budget: 8, min_quality: 0.6 })
+```
+
+`route_verdict` adds the ranked runners-up, the constraint filters (`max_latency_p95_ms`, `budget`, `min_quality`, `require_operational`, `exclude_deprecated`), and the AFTA-signed receipt the agent can audit later. Credits come from [tensorfeed.ai/developers/agent-payments](https://tensorfeed.ai/developers/agent-payments).
+
+### Why it matters
+
+Models, prices, and latency move week to week. `route_verdict` is one signed call an agent can act on now and later prove why it routed the way it did, without rebuilding the comparison from scratch each time.
+
 ## Publish a new version
 
 From the main tensorfeed repo:

package/dist/index.js

@@ -584,6 +584,124 @@ registerTool('premium_routing', 'Get a ranked list of recommended AI models for
         : '';
     return { content: [{ type: 'text', text: `Routing for "${data.task}":\n\n${list}${billing}` }] };
 });
+registerTool('route_verdict_preview', "Free. TensorFeed's signed routing decision that fuses live pricing, contamination-discounted benchmarks, real production usage, MEASURED p95 latency, incident state, and deprecation flags into the single best model for a task or a named model, with the reasoning. For ranked runners-up, constraint filters, and an AFTA-signed receipt you can audit, use route_verdict. 10 calls per day per IP.", {
+    task: z.enum(['code', 'reasoning', 'creative', 'general']).optional().describe('Task type to route for (code, reasoning, creative, general). Provide task or model.'),
+    model: z.string().optional().describe('Model id or display name to narrow the verdict to one model (e.g. "Claude Opus 4.7" or "claude-opus-4-7"). Provide task or model.'),
+}, async ({ task, model }) => {
+    if (!task && !model) {
+        return {
+            content: [
+                {
+                    type: 'text',
+                    text: 'Provide a task (code, reasoning, creative, general) or a model id/name to get a routing verdict.',
+                },
+            ],
+        };
+    }
+    const params = new URLSearchParams();
+    if (task)
+        params.set('task', task);
+    if (model)
+        params.set('model', model);
+    const data = (await fetchJSON(`/preview/route-verdict?${params}`));
+    if (!data.verdict) {
+        return {
+            content: [{ type: 'text', text: `No routing verdict available for that query.\n${data.claim ?? ''}` }],
+        };
+    }
+    const v = data.verdict;
+    const t = data.trust;
+    const fresh = [
+        data.data_freshness.pricing ? `pricing ${data.data_freshness.pricing}` : null,
+        data.data_freshness.probe ? `latency ${data.data_freshness.probe}` : null,
+        data.data_freshness.status ? `status ${data.data_freshness.status}` : null,
+    ]
+        .filter(Boolean)
+        .join(', ');
+    const trustLine = `Trust: usage ${t.usage_corroborated ? 'corroborated' : 'uncorroborated'}, benchmark contamination ${t.benchmark_contamination}, operational ${t.operational_layer}, latency ${t.latency_layer}`;
+    const rl = data.rate_limit ? `\nPreview: ${data.rate_limit.remaining} of ${data.rate_limit.limit} calls left today (${data.rate_limit.scope}).` : '';
+    return {
+        content: [
+            {
+                type: 'text',
+                text: `Route Verdict: ${v.model.name} (${v.model.provider})\n` +
+                    `Why: ${v.why}\n` +
+                    `Blended price: $${v.pricing.blended}/1M tokens (in $${v.pricing.input}, out $${v.pricing.output})\n` +
+                    `${trustLine}${fresh ? `\nData freshness: ${fresh}` : ''}\n` +
+                    `Upgrade: route_verdict (1 credit) adds ranked runners-up, constraint filters, and an AFTA-signed receipt you can audit.${rl}`,
+            },
+        ],
+    };
+});
+// ── Tool: route_verdict (1 credit) ──────────────────────────────────
+// The signed routing decision. Single best model plus ranked runners-up
+// for a task or named model, with constraint filters and an AFTA-signed
+// receipt over the exact inputs. Strict premium, no free trial. The free
+// route_verdict_preview is the no-auth taste of the same engine.
+registerTool('route_verdict', "Costs 1 credit ($0.02). The signed routing decision: TensorFeed's single best model plus ranked runners-up for your task or a named model, fused from live pricing, contamination-discounted benchmarks, real production usage, MEASURED p95 latency, incident state, and deprecation flags, with an AFTA-signed receipt over the exact inputs so you can prove why you routed. Versus the free route_verdict_preview it adds ranked runners-up, constraint filters (max p95 latency, budget, min quality, operational-only, exclude-deprecated), the signed receipt, and no rate limit. Get credits at tensorfeed.ai/developers/agent-payments. Strict premium, no free trial.", {
+    task: z.enum(['code', 'reasoning', 'creative', 'general']).optional().describe('Task type to route for (code, reasoning, creative, general). Provide task or model.'),
+    model: z.string().optional().describe('Model id or display name to narrow the verdict to one model. Provide task or model.'),
+    max_latency_p95_ms: z.number().optional().describe('Drop candidates whose measured p95 latency exceeds this value (ms).'),
+    budget: z.number().optional().describe('Max blended USD per 1M tokens'),
+    min_quality: z.number().min(0).max(1).optional().describe('Minimum trust-discounted quality score in [0, 1]'),
+    require_operational: z.boolean().optional().describe('Default true. Set false to keep candidates known down or in failover.'),
+    exclude_deprecated: z.boolean().optional().describe('Default true. Set false to keep deprecated or sunsetted models.'),
+}, async ({ task, model, max_latency_p95_ms, budget, min_quality, require_operational, exclude_deprecated }) => {
+    if (!task && !model) {
+        return {
+            content: [
+                {
+                    type: 'text',
+                    text: 'Provide a task (code, reasoning, creative, general) or a model id/name to get a signed routing verdict.',
+                },
+            ],
+        };
+    }
+    const params = new URLSearchParams();
+    if (task)
+        params.set('task', task);
+    if (model)
+        params.set('model', model);
+    if (typeof max_latency_p95_ms === 'number')
+        params.set('max_latency_p95_ms', String(max_latency_p95_ms));
+    if (typeof budget === 'number')
+        params.set('budget', String(budget));
+    if (typeof min_quality === 'number')
+        params.set('min_quality', String(min_quality));
+    if (typeof require_operational === 'boolean')
+        params.set('require_operational', String(require_operational));
+    if (typeof exclude_deprecated === 'boolean')
+        params.set('exclude_deprecated', String(exclude_deprecated));
+    const data = (await fetchJSON(`/premium/route-verdict?${params}`, { auth: true }));
+    if (!data.verdict) {
+        return {
+            content: [{ type: 'text', text: `No routing verdict matched your filters.\n${data.claim ?? ''}` }],
+        };
+    }
+    const v = data.verdict;
+    const t = data.trust;
+    const f = data.filters_applied;
+    const runners = (data.runners_up ?? [])
+        .map((r) => `  #${r.rank} ${r.model.name} (${r.model.provider}) blended $${r.pricing.blended}/1M\n     ${r.why}`)
+        .join('\n');
+    const trustLine = `Trust: usage ${t.usage_corroborated ? 'corroborated' : 'uncorroborated'}, benchmark contamination ${t.benchmark_contamination}, operational ${t.operational_layer}, latency ${t.latency_layer}`;
+    const filterLine = `Filters: max p95 ${f.max_latency_p95_ms ?? 'none'}ms, require_operational ${f.require_operational}, exclude_deprecated ${f.exclude_deprecated}`;
+    const billing = data.billing
+        ? `\n\nCharged ${data.billing.credits_charged} credit. Remaining: ${data.billing.credits_remaining}.`
+        : '';
+    return {
+        content: [
+            {
+                type: 'text',
+                text: `Route Verdict: ${v.model.name} (${v.model.provider}) score ${v.composite_score}\n` +
+                    `Why: ${v.why}\n` +
+                    `Blended price: $${v.pricing.blended}/1M tokens (in $${v.pricing.input}, out $${v.pricing.output})\n\n` +
+                    `Runners-up:\n${runners || '  (none)'}\n\n` +
+                    `${trustLine}\n${filterLine}\n\nClaim: ${data.claim}${billing}`,
+            },
+        ],
+    };
+});
 // ── Tool: pricing_series_free (free, 7-day cap) ─────────────────────
 registerTool('pricing_series_free', 'Daily price points for one AI model over the last 1 to 7 days, free. For windows up to 90 days use pricing_series (1 credit).', {
     model: z.string().describe('Model id or display name (e.g. "Claude Opus 4.7" or "claude-opus-4-7")'),

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tensorfeed/mcp-server",
-  "version": "1.36.2",
+  "version": "1.36.4",
   "mcpName": "ai.tensorfeed/mcp-server",
   "description": "MCP server for TensorFeed.ai - AI news, status, model data, and premium endpoints (routing, cost projection, forecast, history series, news search, webhook watches with digest tier, enriched directory) for AI agents",
   "main": "dist/index.js",

package/server.json

@@ -3,7 +3,7 @@
   "name": "ai.tensorfeed/mcp-server",
   "title": "TensorFeed",
   "description": "AI news, model pricing, service status, and machine-payable premium tools. AFTA-certified.",
-  "version": "1.36.2",
+  "version": "1.36.4",
   "websiteUrl": "https://tensorfeed.ai/developers/agent-payments",
   "repository": {
     "url": "https://github.com/RipperMercs/tensorfeed-mcp",
@@ -20,7 +20,7 @@
       "registryType": "npm",
       "registryBaseUrl": "https://registry.npmjs.org",
       "identifier": "@tensorfeed/mcp-server",
-      "version": "1.36.2",
+      "version": "1.36.4",
       "runtimeHint": "npx",
       "transport": {
         "type": "stdio"
@@ -78,7 +78,8 @@
         "get_x402_summary",
         "get_x402_publishers",
         "get_x402_leaderboard",
-        "get_x402_recent"
+        "get_x402_recent",
+        "route_verdict_preview"
       ],
       "premiumTools": [
         "premium_routing",
@@ -103,7 +104,8 @@
         "get_account_usage",
         "premium_ai_company",
         "premium_x402_publisher_receipts",
-        "premium_x402_series"
+        "premium_x402_series",
+        "route_verdict"
       ]
     }
   }