@index9/mcp 6.5.0 → 6.5.2

package/README.md

@@ -2,17 +2,15 @@
 
 Discover, shortlist, compare, cost-model, and live-test 300+ AI models from your editor
 
+> Per-editor install configs (Cursor, VS Code, Claude Code, Codex): https://index9.dev/#install
+
 ## Install
 
 ```bash
 npx -y @index9/mcp@latest
 ```
 
-## OpenRouter API key
-
-Optional: set OPENROUTER_API_KEY in your MCP client config for live test_model calls. dryRun=true works without a key.
-
-Add the key as `OPENROUTER_API_KEY` in the MCP server environment for your client. Example:
+Generic MCP client config:
 
 ```json
 {
@@ -28,25 +26,28 @@ Add the key as `OPENROUTER_API_KEY` in the MCP server environment for your clien
 }
 ```
 
-**Claude Code:** Run `claude mcp add --transport stdio index9 -- npx -y @index9/mcp` or add the same config to .mcp.json / ~/.claude.json.
+Claude Code one-liner: `claude mcp add --transport stdio index9 -- npx -y @index9/mcp`.
+
+## OpenRouter API key
+
+Required for live test_model calls. Set OPENROUTER_API_KEY in your MCP client config. dryRun=true works without it.
 
 ## Tools
 
-- **find_models** — Search and paginate AI models by semantic query or filters
-- **get_models** — Get full model metadata by IDs or aliases (batch, up to 100)
-- **compare_models** — Diff 2-10 models across pricing, context, capabilities, and tokenizer
-- **list_facets** — Enumerate available providers, capabilities, modalities, and tokenizers
-- **test_model** — Run live inference or dry-run cost estimation across up to 10 models (requires OpenRouter API key)
+- **list_facets**: List available providers, capabilities, modalities, and tokenizers
+- **find_models**: Search AI models by semantic query or filters
+- **get_models**: Fetch full model metadata for up to 100 IDs or aliases
+- **compare_models**: Diff 2-10 models across pricing, context, capabilities, and tokenizer
+- **test_model**: Run live inference or dry-run cost estimates on 1-10 models (requires OpenRouter API key)
 
-## Response Metadata
+## Response metadata
 
-All tool responses include an `_index9` object with at least:
+All tool responses include an `_index9` object with:
 
-- `apiBaseUrl` — The API base URL used for the request
+- `apiBaseUrl`: API base URL used for the request
 
 Error responses also include:
 
-- `status` — Upstream HTTP status code
-- `_index9.retryAfterSeconds` — Present when `Retry-After` is returned
-- `_index9.rateLimit` — Present when rate-limit headers are returned (`x-ratelimit-limit`, `x-ratelimit-remaining`, `x-ratelimit-reset`)
-
+- `status`: upstream HTTP status code
+- `_index9.retryAfterSeconds`: present when `Retry-After` is returned
+- `_index9.rateLimit`: present when rate-limit headers are returned (`x-ratelimit-limit`, `x-ratelimit-remaining`, `x-ratelimit-reset`)

package/dist/cli.js

@@ -121,10 +121,21 @@ Key rules:
 - Reasoning-capable models (capabilities includes "reasoning") burn hidden reasoning tokens against \`maxTokens\` before emitting visible text. Leave \`maxTokens\` unset, or set it to at least 2000, when testing reasoning models \u2014 otherwise results may fail with finish_reason=length.
 - Cursors are opaque and tied to query/sort/filters. Reuse the same query/sort/filters when paginating. \`limit\` may change between pages.`;
 var TOOLS = {
+  list_facets: {
+    name: "list_facets",
+    title: "List filter vocabulary",
+    summary: "List available providers, capabilities, modalities, and tokenizers",
+    description: `Return the live vocabulary derived from the current model cache: providers (id-prefixes with model counts), capability flags (with counts), input/output modalities, and tokenizers.
+
+Use this once at the start of a session to learn what filter values find_models will accept, instead of trial-and-error against find_models.
+
+Response: { providers: { id, modelCount }[], capabilities: { id, modelCount }[], modalities: { input: string[], output: string[] }, tokenizers: string[], totalModels: number, updatedAt: string (ISO 8601) }. No inputs.`,
+    requiresKey: false
+  },
   find_models: {
     name: "find_models",
     title: "Search AI models",
-    summary: "Search and paginate AI models by semantic query or filters",
+    summary: "Search AI models by semantic query or filters",
     description: `Filter ${MODEL_COUNT} AI models by structured constraints (capabilities, price, context, modality, provider). The semantic \`q\` is a soft tiebreaker on top of filters \u2014 trust filters first; weak \`q\`-only queries surface as \`meta.confidence: "low"\` with a hint to drop \`q\` and use \`sortBy=created\` or \`sortBy=price\` instead.
 
 Call this after list_facets (or directly when you already know the filters). Always include one call with sortBy=created (no q) when recommending models to the user \u2014 your training-data list of "good" models is likely stale, and sorting purely by price surfaces old tiny models and misses this month's cheap frontier.
@@ -153,7 +164,7 @@ Pass result.id to get_models for full specs or to test_model for live testing.`,
   get_models: {
     name: "get_models",
     title: "Inspect AI models",
-    summary: "Get full model metadata by IDs or aliases (batch, up to 100)",
+    summary: "Fetch full model metadata for up to 100 IDs or aliases",
     description: `Get full specs for one or more models by id or alias. Accepts up to ${LIMITS.getModelsMax} ids per call \u2014 use this for batch comparison.
 
 Call after find_models to inspect candidates, or directly when the user names a model (format: 'provider/model-name').
@@ -189,21 +200,10 @@ Optional: pass \`expectedPromptTokens\` AND \`expectedCompletionTokens\` to also
 Accepts the same alias formats as get_models. Unknown ids are returned in missingIds (with \`suggestions[id]\` as newest-first \`{id, name, created}\` entries when partial matches exist, plus \`missingDiagnostics\` carrying a machine-readable reason per id). When fewer than 2 ids resolve, this returns 400 with the diagnostics so you can retry with \`suggestions[id][0].id\` for each missing id.`,
     requiresKey: false
   },
-  list_facets: {
-    name: "list_facets",
-    title: "List filter vocabulary",
-    summary: "Enumerate available providers, capabilities, modalities, and tokenizers",
-    description: `Return the live vocabulary derived from the current model cache: providers (id-prefixes with model counts), capability flags (with counts), input/output modalities, and tokenizers.
-
-Use this once at the start of a session to learn what filter values find_models will accept, instead of trial-and-error against find_models.
-
-Response: { providers: { id, modelCount }[], capabilities: { id, modelCount }[], modalities: { input: string[], output: string[] }, tokenizers: string[], totalModels: number, updatedAt: string (ISO 8601) }. No inputs.`,
-    requiresKey: false
-  },
   test_model: {
     name: "test_model",
     title: "Test AI models live",
-    summary: "Run live inference or dry-run cost estimation across up to 10 models",
+    summary: "Run live inference or dry-run cost estimates on 1-10 models",
     description: `Run model tests on 1-${LIMITS.testModelsMax} models. Use dryRun=true to estimate token usage/cost, or dryRun=false (default) to run live OpenRouter inference.
 
 When dryRun=true:
@@ -262,13 +262,13 @@ var SITE = {
     installCta: "Install"
   },
   hero: {
-    title: "Pick AI models from live data, not your assistant's memory.",
-    subtitle: "An MCP server with five tools for Claude Code, Cursor, VS Code, and Codex. Search, compare, cost-model, and live-test 300+ models.",
-    pricingNote: "Free. You only pay OpenRouter for live model calls.",
+    title: "Index9 helps coding assistants pick models from live data.",
+    subtitle: "MCP tools for Claude Code, Cursor, VS Code, and Codex. Search, compare, cost-model, and live-test 300+ models.",
+    pricingNote: "Free to install. Live tests use your OpenRouter key.",
     seeHowItWorks: "See a real session",
     updatedBadge: "OpenRouter data \xB7 refreshed ",
     toolList: [
-      { name: "list_facets", role: "vocabulary" },
+      { name: "list_facets", role: "filter vocabulary" },
       { name: "find_models", role: "search & filter" },
       { name: "get_models", role: "full specs" },
       { name: "compare_models", role: "side-by-side diff" },
@@ -285,6 +285,7 @@ var SITE = {
     ]
   },
   howItWorks: {
+    label: "Workflow",
     heading: "How it works",
     subtitle: "Index9 adds 5 tools to your editor. Your assistant calls them when you ask about models.",
     steps: [
@@ -308,7 +309,7 @@ var SITE = {
   caseStudy: {
     label: "Real session",
     heading: "A real session, not a mockup",
-    subheading: "Claude Sonnet 4.6 driven through the index9 MCP server, picking a model for a TypeScript code-review bot. Real prompt, real tool calls, real verdict.",
+    subheading: "Claude Sonnet 4.6 using Index9 to pick a model for a TypeScript code-review bot. Real prompt, tool calls, and verdict.",
     prompt: {
       title: "The prompt",
       body: "Pick a model for a TypeScript code-review bot that runs on every PR. I want real quality without paying frontier rates. Quote model ids verbatim from the tool responses."
@@ -339,44 +340,44 @@ var SITE = {
         }
       ]
     },
-    consideredTitle: "Candidates the session evaluated",
-    consideredSubtitle: "Lifted from the tool responses. Per-PR cost is the dry-run test_model output for a 3,000 token diff and an 800 token review.",
+    consideredTitle: "Candidates evaluated",
+    consideredSubtitle: "5 evaluated, 1 pick",
     consideredRows: [
       {
         id: "anthropic/claude-opus-4.7-fast",
         age: "5d ago",
         decision: "skip",
-        reason: "$30/M input, filtered out by the maxPrice constraint on the second find_models call"
+        reason: "$30/M input, above the maxPrice filter"
       },
       {
         id: "mistralai/devstral-medium",
         age: "10mo ago",
         decision: "tested",
-        reason: "$0.40/M input, completion-heavy at $2/M, $0.0028 per PR"
+        reason: "$0.40/M input, $2/M output, $0.0028 per PR"
       },
       {
         id: "qwen/qwen3-coder",
         age: "10mo ago",
         decision: "tested",
-        reason: "$0.22/M input, 1M context but no file_input, $0.0021 per PR"
+        reason: "$0.22/M input, 1M context, no file_input, $0.0021 per PR"
       },
       {
         id: "google/gemini-3.1-flash-lite",
         age: "1w ago",
         decision: "tested",
-        reason: "$0.25/M input, newest of the shortlist, adds reasoning and vision, $0.00195 per PR"
+        reason: "$0.25/M input, newest finalist, reasoning + vision, $0.00195 per PR"
       },
       {
         id: "mistralai/codestral-2508",
         age: "9mo ago",
         decision: "shortlisted",
-        reason: "$0.30/M input, Mistral's coding-specific model, $0.00162 per PR"
+        reason: "$0.30/M input, coding-specific, $0.00162 per PR"
       }
     ],
     verdict: {
       title: "The pick",
       model: "mistralai/codestral-2508",
-      body: "Lowest realistic per-PR cost at $0.00162 on a 3,000 token diff and an 800 token review. Mistral's coding-specific model, structured output, file input for diffs, 256K context. About 100\xD7 cheaper on input than running claude-opus-4.7-fast on every commit."
+      body: "$0.00162 per PR (3K diff + 800 review, dry-run). Coding-specific, structured output, file input, 256K context. About 100\xD7 cheaper on input than claude-opus-4.7-fast."
     }
   },
   toolsSection: {
@@ -442,7 +443,7 @@ var SITE = {
     items: [
       {
         question: "What is MCP?",
-        answer: "A protocol that lets AI assistants call external tools. Index9 is one of those tools.",
+        answer: "A protocol that lets AI assistants call external tools. Index9 is one.",
         link: {
           label: "Learn more about MCP",
           url: "https://modelcontextprotocol.io"
@@ -450,45 +451,45 @@ var SITE = {
       },
       {
         question: "Who is index9 for?",
-        answer: "Developers using Claude Code, Cursor, VS Code, or Codex who want their assistant to pick models from current data instead of training-data memory.",
+        answer: "Developers in Claude Code, Cursor, VS Code, or Codex who want model picks from live data, not training memory.",
         link: null
       },
       {
         question: "Does it pick the model for me?",
-        answer: "No. It gives your assistant the data: search results, specs, cost diffs, live test outputs. Your assistant makes the call.",
+        answer: "No. It hands your assistant the data: search, specs, cost diffs, test outputs. The assistant makes the call.",
         link: null
       },
       {
         question: "How does live testing work?",
-        answer: `test_model sends your prompt to up to ${LIMITS.testModelsMax} models via OpenRouter and returns output, latency, tokens, and cost. Dry-run mode estimates cost without running inference.`,
+        answer: `test_model runs your prompt against up to ${LIMITS.testModelsMax} models via OpenRouter and returns output, latency, tokens, and cost. Set dryRun=true to estimate cost without inference.`,
         link: null
       },
       {
         question: "Which models?",
-        answer: `${MODEL_COUNT} from OpenRouter: OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, and more. Metadata refreshes every 30 minutes.`,
+        answer: `${MODEL_COUNT} from OpenRouter: OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek. Metadata refreshes every 30 minutes.`,
         link: null
       },
       {
         question: "Do you store my prompts or keys?",
-        answer: "No. Index9 doesn't store prompts, outputs, or API keys. Live tests are proxied straight to OpenRouter.",
+        answer: "No. Index9 stores no prompts, outputs, or API keys. Live tests proxy directly to OpenRouter.",
         link: null
       },
       {
         question: "What's the project status?",
-        answer: "Stable. Issues and feature requests on GitHub.",
+        answer: "Stable. File issues and feature requests on GitHub.",
         link: null
       }
     ]
   },
   install: {
     label: "Setup",
-    heading: "Install",
-    subheading: "Pick your editor and paste the config.",
+    heading: "Install Index9",
+    subheading: "Choose a client, copy the config, restart your assistant.",
     configs: [
       {
         id: "cursor",
         label: "Cursor",
-        paths: [".cursor/mcp.json"],
+        paths: [".cursor/mcp.json", "~/.cursor/mcp.json"],
         config: `{
   "mcpServers": {
     "index9": {
@@ -501,7 +502,7 @@ var SITE = {
       {
         id: "vscode",
         label: "VS Code",
-        paths: [".vscode/mcp.json"],
+        paths: [".vscode/mcp.json", "user profile mcp.json"],
         config: `{
   "servers": {
     "index9": {
@@ -516,7 +517,7 @@ var SITE = {
         id: "claude-code",
         label: "Claude Code",
         paths: [],
-        config: `claude mcp add --transport stdio index9 -- npx -y @index9/mcp@latest`,
+        config: `claude mcp add --transport stdio --scope user index9 -- npx -y @index9/mcp@latest`,
         copyHint: "# Run in terminal (adds to ~/.claude.json)"
       },
       {
@@ -525,19 +526,33 @@ var SITE = {
         paths: [],
         config: `codex mcp add index9 -- npx -y @index9/mcp@latest`,
         copyHint: "# Run in terminal (adds to ~/.codex/config.toml)"
+      },
+      {
+        id: "generic",
+        label: "Any MCP client",
+        paths: ["mcp.json"],
+        config: `{
+  "mcpServers": {
+    "index9": {
+      "command": "npx",
+      "args": ["-y", "@index9/mcp@latest"]
+    }
+  }
+}`
       }
     ],
     copyButton: "Copy",
     copiedButton: "Copied",
     copyAriaLabel: "Copy configuration",
     copiedAnnouncement: "Configuration copied to clipboard",
-    setupNote: "Want live tests?",
-    setupLink: "Add an OpenRouter key",
+    nextStep: "Search, compare, and dry-run cost estimates work without a key.",
+    setupNote: "Live tests need",
+    setupLink: "an OpenRouter key",
     setupUrl: "https://github.com/index9-org/mcp#openrouter-api-key"
   },
   footer: {
     brand: "index9",
-    tagline: "MCP-native model picker. Live data from OpenRouter.",
+    tagline: "Live model data, via MCP.",
     copyrightSuffix: "index9",
     github: "GitHub",
     privacy: "Privacy",
@@ -547,7 +562,8 @@ var SITE = {
 var README = {
   mcpDescription: `Discover, shortlist, compare, cost-model, and live-test ${MODEL_COUNT} AI models from your editor`,
   mcpInstall: {
-    envNote: "Optional: set OPENROUTER_API_KEY in your MCP client config for live test_model calls. dryRun=true works without a key."
+    envNote: "Required for live test_model calls. Set OPENROUTER_API_KEY in your MCP client config. dryRun=true works without it.",
+    envVarDescription: "Required for live test_model calls. dryRun=true works without it."
   }
 };
 

package/manifest.json

@@ -1,7 +1,7 @@
 {
   "manifest_version": "0.3",
   "name": "index9",
-  "version": "6.4.0",
+  "version": "6.5.1",
   "description": "Discover, shortlist, compare, cost-model, and live-test 300+ AI models from your editor",
   "author": {
     "name": "Index9"
@@ -11,42 +11,36 @@
     "entry_point": "dist/cli.js",
     "mcp_config": {
       "command": "node",
-      "args": [
-        "${__dirname}/dist/cli.js"
-      ],
+      "args": ["${__dirname}/dist/cli.js"],
       "env": {
         "OPENROUTER_API_KEY": "${user_config.openrouter_api_key}"
       }
     }
   },
   "tools": [
+    {
+      "name": "list_facets",
+      "description": "List available providers, capabilities, modalities, and tokenizers"
+    },
     {
       "name": "find_models",
-      "description": "Search and paginate AI models by semantic query or filters"
+      "description": "Search AI models by semantic query or filters"
     },
     {
       "name": "get_models",
-      "description": "Get full model metadata by IDs or aliases (batch, up to 100)"
+      "description": "Fetch full model metadata for up to 100 IDs or aliases"
     },
     {
       "name": "compare_models",
       "description": "Diff 2-10 models across pricing, context, capabilities, and tokenizer"
     },
-    {
-      "name": "list_facets",
-      "description": "Enumerate available providers, capabilities, modalities, and tokenizers"
-    },
     {
       "name": "test_model",
-      "description": "Run live inference or dry-run cost estimation across up to 10 models"
+      "description": "Run live inference or dry-run cost estimates on 1-10 models"
     }
   ],
   "compatibility": {
-    "platforms": [
-      "darwin",
-      "linux",
-      "win32"
-    ],
+    "platforms": ["darwin", "linux", "win32"],
     "runtimes": {
       "node": ">=20.0.0"
     }
@@ -55,7 +49,7 @@
     "openrouter_api_key": {
       "type": "string",
       "title": "OpenRouter API Key",
-      "description": "Required for live test_model calls. dryRun=true does not require it.",
+      "description": "Required for live test_model calls. dryRun=true works without it.",
       "sensitive": true,
       "required": false
     }

package/package.json

@@ -1,6 +1,7 @@
 {
   "name": "@index9/mcp",
-  "version": "6.5.0",
+  "version": "6.5.2",
+  "description": "Discover, shortlist, compare, cost-model, and live-test 300+ AI models from your editor",
   "license": "MIT",
   "repository": {
     "type": "git",
@@ -11,7 +12,8 @@
   },
   "files": [
     "dist",
-    "manifest.json"
+    "manifest.json",
+    "server.json"
   ],
   "type": "module",
   "main": "./dist/cli.js",
@@ -33,6 +35,7 @@
   "engines": {
     "node": ">=20"
   },
+  "mcpName": "io.github.index9-org/mcp",
   "scripts": {
     "build": "node ../../scripts/generate-docs.mjs && tsup",
     "clean": "rm -rf dist",

package/server.json

@@ -0,0 +1,31 @@
+{
+  "$schema": "https://static.modelcontextprotocol.io/schemas/2025-12-11/server.schema.json",
+  "name": "io.github.index9-org/mcp",
+  "title": "Index9",
+  "description": "Discover, shortlist, compare, cost-model, and live-test 300+ AI models from your editor",
+  "repository": {
+    "url": "https://github.com/index9-org/mcp",
+    "source": "github"
+  },
+  "version": "6.5.1",
+  "packages": [
+    {
+      "registryType": "npm",
+      "registryBaseUrl": "https://registry.npmjs.org",
+      "identifier": "@index9/mcp",
+      "version": "6.5.1",
+      "transport": {
+        "type": "stdio"
+      },
+      "environmentVariables": [
+        {
+          "name": "OPENROUTER_API_KEY",
+          "description": "Required for live test_model calls. dryRun=true works without it.",
+          "format": "string",
+          "isRequired": false,
+          "isSecret": true
+        }
+      ]
+    }
+  ]
+}