@intangle/mcp-server 2.6.6 → 2.6.8

package/README.md

@@ -1,23 +1,23 @@
 # @intangle/mcp-server
 
-Model Context Protocol server for [Intangle](https://intangle.app) - Intelligent AI context that persists across conversations.
+Model Context Protocol server for [Intangle](https://renvoi.co/intangle) - Intelligent AI context that persists across conversations.
 
 ## What is Intangle?
 
 Intangle is an AI context system that gives Claude and other AI assistants persistent, searchable context across all your conversations and isolated spaces. Unlike traditional chat interfaces where context is lost between sessions, Intangle maintains your personal context across isolated spaces, with intelligent storing and retrieval.
 
-**Important: You need an Intangle account to use this MCP server. Sign up at [intangle.app](https://intangle.app)**
+**Important: You need an Intangle account to use this MCP server. Sign up at [renvoi.co/intangle](https://renvoi.co/intangle)**
 
 ## Setup
 
-1. **Create an Intangle account** at [intangle.app](https://intangle.app) and get your API key from Settings → Connect
+1. **Create an Intangle account** at [renvoi.co/intangle](https://renvoi.co/intangle) and get your API key from Settings → Connect
 
 2. **Configure Provider Access ** 
 
 Claude code by running the command below (recommend "-s" user for global across projects):
 
 ```bash
-claude mcp add intangle -s user --env MCP_API_KEY=<your-api-key-here> --env NEXT_APP_URL=https://app.intangle.app -- npx -y @intangle/mcp-server
+claude mcp add intangle -s user --env MCP_API_KEY=<your-api-key-here> --env NEXT_APP_URL=https://renvoi.co/intangle -- npx -y @intangle/mcp-server
 ```
 
 If your deployment is behind Vercel Deployment Protection (for example staging), add:
@@ -36,7 +36,7 @@ Standard MCP (Claude desktop, Cursor): by adding to your `claude_desktop_config.
       "args": ["@intangle/mcp-server"],
       "env": {
         "MCP_API_KEY": "your-api-key-here",
-        "NEXT_APP_URL": "https://app.intangle.app",
+        "NEXT_APP_URL": "https://renvoi.co/intangle",
         "VERCEL_BYPASS_TOKEN": "optional-vercel-bypass-token-for-protected-deployments"
       }
     }
@@ -92,7 +92,7 @@ If Claude can't access your context:
 
 ## Support
 
-- Website: [intangle.app](https://intangle.app)
+- Website: [renvoi.co/intangle](https://renvoi.co/intangle)
 - Issues: [github.com/intangle/mcp-server/issues](https://github.com/intangle/mcp-server/issues)
 - Email: hello@intangle.app
 

package/dist/index.js

@@ -2,9 +2,9 @@
 /**
  * Intangle MCP Server - Exposes Intangle context functionality through MCP
  */
-import { readFileSync } from "fs";
-import { dirname, join } from "path";
-import { fileURLToPath } from "url";
+import { appendFileSync, readFileSync } from "node:fs";
+import { dirname, join } from "node:path";
+import { fileURLToPath } from "node:url";
 import { Server } from "@modelcontextprotocol/sdk/server/index.js";
 import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
 import { CallToolRequestSchema, ErrorCode, ListResourceTemplatesRequestSchema, ListResourcesRequestSchema, ListToolsRequestSchema, McpError, ReadResourceRequestSchema, } from "@modelcontextprotocol/sdk/types.js";
@@ -13,7 +13,6 @@ import fetch from "node-fetch";
 import { TOOLS } from "./tool-definitions.js";
 // ... imports ...
 // Debug logging
-import { appendFileSync } from "fs";
 const DEBUG_LOG = "/tmp/intangle-mcp-debug.log";
 function log(msg) {
     try {
@@ -30,7 +29,7 @@ try {
     config({ quiet: true });
     config({ path: ".env.local", quiet: true });
     // Base URL for API calls to Next.js app
-    const API_BASE_URL = process.env.NEXT_APP_URL || "https://app.intangle.app";
+    const API_BASE_URL = process.env.NEXT_APP_URL || "https://renvoi.co/intangle";
     const MCP_API_KEY = process.env.MCP_API_KEY;
     const VERCEL_BYPASS_TOKEN = process.env.VERCEL_BYPASS_TOKEN || process.env.VERCEL_PROTECTION_BYPASS;
     log(`CWD: ${process.cwd()}`);
@@ -158,7 +157,7 @@ try {
             if (contentType.includes("text/html") ||
                 /^\s*<!DOCTYPE\s+html/i.test(responseBody) ||
                 /^\s*<html\b/i.test(responseBody)) {
-                throw new Error(`API call to ${buildApiUrl(endpoint)} returned HTML instead of JSON. Check NEXT_APP_URL and point it at the app host, e.g. https://app.intangle.app.`);
+                throw new Error(`API call to ${buildApiUrl(endpoint)} returned HTML instead of JSON. Check NEXT_APP_URL and point it at the app host, e.g. https://renvoi.co/intangle.`);
             }
             return JSON.parse(responseBody);
         }
@@ -399,10 +398,6 @@ try {
         const { space_id } = args;
         return makeApiCall("view-space", args);
     }
-    async function handleStart(args) {
-        const { space_id } = args;
-        return makeApiCall("start", args);
-    }
     async function handleUpdateSpace(args) {
         if (!args.space_id) {
             throw new Error("space_id is required. Use view_spaces to see available options.");
@@ -529,8 +524,7 @@ try {
                     result = await handleViewSpace(args);
                     break;
                 case "start":
-                    result = await handleStart(args);
-                    break;
+                    throw new McpError(ErrorCode.InvalidRequest, "The start tool is disabled because its curated briefing can be stale or misleading. Use view_spaces, view_space, search, fetch_items, message, and update_memory instead.");
                 case "update_memory":
                 case "update_space":
                     result = await handleUpdateSpace(args);

package/dist/tool-definitions.js

@@ -39,7 +39,7 @@ const TOOL_DEFINITIONS = [
     {
         name: "fetch_items",
         title: "Fetch Items by ID",
-        description: "Retrieve full content for specific items by ID, including `artifact_*` and `summary_*` ids. Artifact fetches return content plus render metadata and `config`/`state`/`data` when available.",
+        description: "Retrieve full content for specific items by ID, including `file_*`, `artifact_*`, and `summary_*` ids. File fetches return extracted content, file metadata, chunk metadata, and decrypted `data.source_data_url` when available; use this before asking Tan to reason over a specific uploaded contract/file in detail. Artifact fetches return content plus render metadata and `config`/`state`/`data` when available.",
         inputSchema: {
             type: "object",
             properties: {
@@ -49,12 +49,12 @@ const TOOL_DEFINITIONS = [
                 },
                 id: {
                     type: "string",
-                    description: "Single ID to fetch (e.g. `mem_123`, `task_456`, `skill_789`, `doc_321`, `artifact_654`, or `summary_987`)."
+                    description: "Single ID to fetch (e.g. `mem_123`, `task_456`, `skill_789`, `doc_321`, `file_654`, `artifact_654`, or `summary_987`)."
                 },
                 ids: {
                     type: "array",
                     items: { type: "string" },
-                    description: "Array of IDs to fetch (mix of context, task, skill, document, artifact, insight, or conversation IDs). Use this to fetch multiple items in one call."
+                    description: "Array of IDs to fetch (mix of context, task, skill, document, file, artifact, insight, or conversation IDs). Use this to fetch multiple items in one call."
                 }
             }
         }
@@ -426,21 +426,6 @@ const TOOL_DEFINITIONS = [
             required: ["space_id", "project_id"]
         }
     },
-    {
-        name: "start",
-        title: "Start Space Session",
-        description: "Load a space fast for a new external caller. Returns assistant-curated current items, useful insights, skills, recent conversations, and assistant preferences so the caller can get oriented quickly and fetch deeper details only when needed. Use this before asking Tan to do work in that space.",
-        inputSchema: {
-            type: "object",
-            properties: {
-                space_id: {
-                    type: "string",
-                    description: "Space to start (e.g., 'renvoi', 'personal')"
-                }
-            },
-            required: ["space_id"]
-        }
-    },
     {
         name: "view_spaces",
         title: "View Spaces",
@@ -468,11 +453,11 @@ const TOOL_DEFINITIONS = [
                 config_badges: {
                     type: "array",
                     items: { type: "string" },
-                    description: "Array of keywords for start tool keywords in this NEW space. These keywords help steer what context gets loaded when running 'start' for this space. Examples: ['fitness', 'recipes', 'travel planning'], ['client work', 'meetings', 'proposals'], ['TypeScript', 'React', 'API design']. They act as search terms to prioritize relevant context."
+                    description: "Array of keywords for this NEW space. Examples: ['fitness', 'recipes', 'travel planning'], ['client work', 'meetings', 'proposals'], ['TypeScript', 'React', 'API design']. They help agents understand and prioritize the space."
                 },
                 startup_preferences: {
                     type: "string",
-                    description: "AI behavior preferences for this NEW space. Guides how AI assistants and agents should communicate and behave (e.g., 'be concise and frank', 'think critically', 'play devils advocate'). This appears in the start briefing under 'Assistant Preferences'."
+                    description: "AI behavior preferences for this NEW space. Guides how AI assistants and agents should communicate and behave (e.g., 'be concise and frank', 'think critically', 'play devils advocate')."
                 }
             },
             required: ["name"]
@@ -506,7 +491,7 @@ const TOOL_DEFINITIONS = [
                 },
                 project_id: {
                     type: "string",
-                    description: "Optional: Project to add items to. All items created in this request will be linked to this project and appear on its canvas. Get project IDs from the start tool output."
+                    description: "Optional: Project to add items to. All items created in this request will be linked to this project and appear on its canvas. Get project IDs from view_projects or search."
                 },
                 group_id: {
                     type: "string",
@@ -676,7 +661,6 @@ const TOOL_DEFINITIONS = [
                                 "insights_user",
                                 "agents",
                                 "skills",
-                                "conversations",
                                 "dashboard"
                             ],
                             description: "Optional: filter folders by sidebar section."
@@ -699,7 +683,6 @@ const TOOL_DEFINITIONS = [
                                     "insights_user",
                                     "agents",
                                     "skills",
-                                    "conversations",
                                     "dashboard"
                                 ],
                                 description: "Sidebar section this folder belongs to."
@@ -751,7 +734,7 @@ const TOOL_DEFINITIONS = [
     {
         name: "message",
         title: "Message Assistant",
-        description: "Primary way to use Intangle for memory, communicating and collaborating with Tan agent in a shared thread. Send a request and either wait for one assistant turn to complete or submit asynchronously for later polling. External agents should include a stable participant_id, participant_type=`agent`, and participant_label so Tan can identify who is speaking and keep progress updates attached to the right caller. Returns assistant text plus continuity IDs (`session_id`, `conversation_id`) so callers can continue the same thread reliably across turns.",
+        description: "Primary way to use Intangle for memory, communicating and collaborating with Tan agent in a shared thread. Send a request and either wait for one assistant turn to complete or submit asynchronously for later polling. External agents should include a stable participant_id, participant_type=`agent`, and participant_label so Tan can identify who is speaking and keep progress updates attached to the right caller. Conversations are backed by `summary_*` IDs; reuse the returned `conversation_id` and reuse the returned `session_id` so callers can continue the same thread reliably across turns. For shared coordination channels, pass a stable `channel_id`; channel and project/group coordinates scope context, while the stable idempotency key anchors retries to one durable turn and the returned conversation_id remains the durable thread anchor.",
         inputSchema: {
             type: "object",
             properties: {
@@ -773,27 +756,44 @@ const TOOL_DEFINITIONS = [
                 },
                 new_conversation: {
                     type: "boolean",
-                    description: "Optional. When true, force a fresh conversation instead of reusing the current active thread."
+                    description: "Optional escape hatch. When true, force a fresh conversation instead of reusing the current active thread."
                 },
                 session_id: {
                     type: "string",
-                    description: "Optional runtime session ID to reuse across calls. Reuse the returned session_id for best continuity and lower warm-up overhead."
+                    description: "Optional runtime session ID to reuse across calls. For follow-up turns, reuse the exact session_id returned by the previous call for best continuity and lower warm-up overhead."
                 },
                 project_id: {
                     type: "string",
                     description: "Optional project scope for the assistant turn."
                 },
+                group_id: {
+                    type: "string",
+                    description: "Optional group scope for the assistant turn. Use this with channel_id for group-scoped coordination."
+                },
+                channel_id: {
+                    type: "string",
+                    description: "Optional route/channel scope for the assistant turn. Defaults to root when omitted. Examples: root, group|legal, project|work/launch."
+                },
                 organization_id: {
                     type: "string",
                     description: "Optional Clerk org ID when the space belongs to an organization graph."
                 },
                 timeout_ms: {
                     type: "integer",
-                    description: "Optional timeout in milliseconds (10,000 to 300,000; default 120,000)."
+                    description: "Optional bounded synchronous wait in milliseconds (10,000 to 45,000; default 45,000 when wait_for_response is true)."
                 },
                 wait_for_response: {
                     type: "boolean",
-                    description: "Optional. Defaults to true. Set false to submit the turn asynchronously when you want a quick acknowledgement first and then poll later with list_conversation_inbox or fetch_conversation_messages."
+                    description: "Optional. Defaults to false so the request returns a durable acknowledgement before infrastructure timeouts. Set true only for a bounded wait; a still-running turn returns status continuing and remains pollable."
+                },
+                client_turn_id: {
+                    type: "string",
+                    description: "Optional caller correlation value. Phoenix derives the durable turn identity from idempotency_key and scopes it by the verified user, graph, space, channel, project/group, and stable retry key."
+                },
+                idempotency_key: {
+                    type: "string",
+                    pattern: "\\S",
+                    description: "REQUIRED stable retry key for this logical message within its room. Reuse it exactly on retry; Phoenix atomically claims one dispatch and replays the original session_id, real conversation_id, and durable result state."
                 },
                 participant_id: {
                     type: "string",
@@ -812,7 +812,7 @@ const TOOL_DEFINITIONS = [
                     description: "Optional IANA timezone for assistant context."
                 }
             },
-            required: ["space_id", "content"]
+            required: ["space_id", "content", "idempotency_key"]
         }
     }
     // DISABLED: memory_action tool is broken and causing errors.
@@ -846,7 +846,6 @@ const READ_ONLY_TOOLS = new Set([
     "usage_status",
     "fetch_conversation_messages",
     "list_conversation_inbox",
-    "start",
     "view_spaces",
     "view_space"
 ]);
@@ -862,7 +861,6 @@ const IDEMPOTENT_TOOLS = new Set([
     "usage_status",
     "fetch_conversation_messages",
     "list_conversation_inbox",
-    "start",
     "view_spaces",
     "view_space"
 ]);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@intangle/mcp-server",
-  "version": "2.6.6",
+  "version": "2.6.8",
   "description": "Model Context Protocol server for Intangle - AI context that persists across conversations",
   "main": "dist/index.js",
   "type": "module",
@@ -10,12 +10,17 @@
     "prebuild": "cp ../web/src/app/api/mcp-remote/tool-definitions.ts ./tool-definitions.ts",
     "build": "tsc",
     "lint": "biome check .",
+    "lint:release": "biome check --config-path=biome.release.json package.json index.ts tool-definitions.ts",
     "lint:fix": "biome check --fix .",
     "prepublishOnly": "npm run build"
   },
   "bin": {
     "intangle-mcp": "dist/index.js"
   },
+  "files": [
+    "dist",
+    "README.md"
+  ],
   "keywords": [
     "mcp",
     "model-context-protocol",
@@ -34,7 +39,7 @@
     "type": "git",
     "url": "git+https://github.com/intangle/mcp-server.git"
   },
-  "homepage": "https://intangle.app",
+  "homepage": "https://renvoi.co/intangle",
   "engines": {
     "node": ">=18.0.0"
   },
@@ -49,5 +54,6 @@
     "@types/node-fetch": "^2.6.12",
     "tsx": "^4.0.0",
     "typescript": "^5.0.0"
-  }
+  },
+  "intangleSourceHash": "8ab73509fba985fb9fa892428eb8149a981d2b044505f85a01a05f054c788c92"
 }

package/.env.local

@@ -1,3 +0,0 @@
-MCP_API_KEY=mcp_placeholder_key_will_be_replaced_by_user_specific_key
-NEXT_APP_URL=https://intangle.app
-VERCEL_BYPASS_TOKEN=igXBygPO1T5dWLBaVyQiN9QB7IECZ50O

package/biome.json

@@ -1,39 +0,0 @@
-{
-  "$schema": "https://biomejs.dev/schemas/1.9.4/schema.json",
-  "vcs": {
-    "enabled": false,
-    "clientKind": "git",
-    "useIgnoreFile": false
-  },
-  "files": {
-    "ignoreUnknown": false
-  },
-  "formatter": {
-    "enabled": true,
-    "formatWithErrors": false,
-    "indentStyle": "space",
-    "indentWidth": 2,
-    "lineEnding": "lf",
-    "lineWidth": 80,
-    "attributePosition": "auto"
-  },
-  "linter": {
-    "enabled": true,
-    "rules": {
-      "recommended": true
-    }
-  },
-  "javascript": {
-    "formatter": {
-      "jsxQuoteStyle": "double",
-      "quoteProperties": "asNeeded",
-      "trailingCommas": "es5",
-      "semicolons": "asNeeded",
-      "arrowParentheses": "always",
-      "bracketSpacing": true,
-      "bracketSameLine": false,
-      "quoteStyle": "double",
-      "attributePosition": "auto"
-    }
-  }
-}