@frontlinesystems/rason-mcp-server 2026.5.3 → 2026.5.5

package/README.md

@@ -2,11 +2,14 @@
 
 # <img src="https://raw.githubusercontent.com/FrontlineSystems/media/refs/heads/main/mcp-128-128-height-reduced.png" alt="Frontline Systems" height="64"> RASON MCP Server
 
-[![RASON MCP](https://img.shields.io/badge/marketplace-v2026.5.1-blue?logo=visualstudiocode&label=RASON%20MCP)](https://marketplace.visualstudio.com/items?itemName=FrontlineSystems.rason-mcp)
+[![RASON MCP (VS Code)](https://vsmarketplacebadges.dev/version/FrontlineSystems.rason-mcp.svg?label=RASON%20MCP%20%28VS%20Code%29)](https://marketplace.visualstudio.com/items?itemName=FrontlineSystems.rason-mcp)
+[![RASON MCP (MCPB)](https://img.shields.io/github/v/release/FrontlineSystems/rason-mcp-server?label=RASON%20MCP%20%28MCPB%29)](https://github.com/FrontlineSystems/rason-mcp-server/releases/latest)
 [![RASON MCP (npm)](https://img.shields.io/npm/v/@frontlinesystems/rason-mcp-server?label=RASON%20MCP%20%28npm%29)](https://www.npmjs.com/package/@frontlinesystems/rason-mcp-server)
 
 AI-powered platform for decision intelligence: building, analyzing, and solving optimization, simulation, data science, and decision models with RASON.
 
+> **Using VS Code?** Check out [**RASON Desktop**](https://www.solver.com/rason-desktop), Frontline's flagship VS Code extension. It bundles this MCP server; a RASON language server (IntelliSense, validation, signature help); built-in local & cloud solvers; model and examples sidebars; live Power BI / Excel integration with auto-refreshing charts; one-click result export; and more.
+
 <sub><i>RASON MCP Server in Claude Desktop — solving a product mix optimization model with results table, constraint utilization chart, and key insights.</i></sub>
 
 <sub><i>🔍 Click the screenshot to view full size.</i></sub>
@@ -97,12 +100,15 @@ This MCP server is more than a REST API wrapper — it's a complete AI infrastru
 
 ## Getting Started
 
-Prerequisites:
+Three ways to install.
+
+### 1. Claude Desktop: drop-in extension
+
+Grab the latest [MCPB release](https://github.com/FrontlineSystems/rason-mcp-server/releases/latest), then in Claude Desktop open **Settings → Extensions → Advanced Settings → Install Extension** and pick the file. On macOS you can also drag the `.mcpb` onto the app window. The install dialog offers to set a RASON bearer token — only needed to solve models or manage them in the cloud, and can be set later from the extension's settings.
 
-- Node.js 18 or later
-- A RASON bearer token only if you want to solve models, run diagnostics, or manage cloud models
+### 2. npm: any MCP client
 
-Add the server to your MCP client's configuration. For most clients, use this server entry under the client's MCP server map:
+For Claude Desktop, Cursor, Windsurf, Claude Code, or any other stdio MCP client, add this entry to your client's MCP config.
 
 ```json
 {
@@ -118,9 +124,7 @@ Add the server to your MCP client's configuration. For most clients, use this se
 }
 ```
 
-You can omit `env` or `RASON_BEARER_TOKEN` to use the no-account tools only.
-
-Where to put this depends on your client:
+Omit `env` or `RASON_BEARER_TOKEN` to use the no-account tools only.
 
 <details>
 <summary><strong>Claude Desktop</strong></summary>
@@ -153,8 +157,6 @@ Create or update an MCP configuration file such as `.vscode/mcp.json` in your wo
 }
 ```
 
-Or install the [RASON MCP Server extension](https://marketplace.visualstudio.com/items?itemName=FrontlineSystems.rason-mcp) for automatic setup.
-
 See [VS Code MCP setup](https://code.visualstudio.com/docs/copilot/customization/mcp-servers) for details.
 
 </details>
@@ -201,6 +203,15 @@ This server works with any MCP-compatible client over stdio. Configure it with:
 
 </details>
 
+### 3. VS Code: one-click extension
+
+Two extensions are available on the Marketplace:
+
+- [**RASON MCP Server**](https://marketplace.visualstudio.com/items?itemName=FrontlineSystems.rason-mcp): this MCP server, standalone.
+- [**RASON Desktop**](https://www.solver.com/rason-desktop): Frontline's flagship VS Code extension. Bundles this MCP server; a RASON language server (IntelliSense, validation, signature help); built-in local & cloud solvers; model and examples sidebars; live Power BI / Excel integration with auto-refreshing charts; one-click result export; and more.
+
+Set the auth token in either extension via the Command Palette: **RASON: Set Auth Token**.
+
 ## Works Without a RASON Account
 
 Example search, model templates, model analysis, and RASON info tools work without credentials. You can explore the full example library and build models before signing up. A [RASON account](https://rason.com) is needed only for solving models and managing models in the cloud.
@@ -224,38 +235,38 @@ Set these environment variables in your MCP client config (see [Getting Started]
 
 ### Examples & Templates
 
-- **search_examples** — Search ~200 example models by keyword or type (optimization, simulation, datamining/data science, calculation/decision tables, flow/multi-stage pipelines)
-- **get_example** — Retrieve a specific example model with full definition and metadata
-- **get_model_template** — Get a bare JSON scaffold for a model type: blank, optimization, simulation, sim-optimization, data-science, workflow, decision-table, power-bi
+- **search_examples**: Search ~200 example models by keyword or type (optimization, simulation, datamining/data science, calculation/decision tables, flow/multi-stage pipelines)
+- **get_example**: Retrieve a specific example model with full definition and metadata
+- **get_model_template**: Get a bare JSON scaffold for a model type: blank, optimization, simulation, sim-optimization, data-science, workflow, decision-table, power-bi
 
 ### Analysis
 
-- **describe_model** — Analyze model structure: type/subtype detection, component inventory, issue identification
-- **about_rason** — RASON capabilities, model types, Frontline Systems ecosystem, and server info
+- **describe_model**: Analyze model structure: type/subtype detection, component inventory, issue identification
+- **about_rason**: RASON capabilities, model types, Frontline Systems ecosystem, and server info
 
 ### Solving
 
-- **solve_model** — Solve a model synchronously (inline RASON, RASON files, or Excel workbooks)
-- **submit_solve** — Submit a model for asynchronous solving (long-running models)
-- **get_solve_status** — Check solving status and progress
-- **get_solve_result** — Retrieve the result once solving is complete
-- **stop_solve** — Stop a running solve (intermediate results may be available)
+- **solve_model**: Solve a model synchronously (inline RASON, RASON files, or Excel workbooks)
+- **submit_solve**: Submit a model for asynchronous solving (long-running models)
+- **get_solve_status**: Check solving status and progress
+- **get_solve_result**: Retrieve the result once solving is complete
+- **stop_solve**: Stop a running solve (intermediate results may be available)
 
 ### Diagnostics
 
-- **diagnose_model** — Run diagnostics synchronously (useful for infeasible/unbounded optimization models)
-- **submit_diagnose** — Submit for asynchronous diagnostics
+- **diagnose_model**: Run diagnostics synchronously (useful for infeasible/unbounded optimization models)
+- **submit_diagnose**: Submit for asynchronous diagnostics
 
 ### Model Management
 
-- **list_models** — List models in your account with optional filters by type and kind
-- **get_model** — Retrieve a model definition (RASON JSON)
-- **post_model** — Create a new model from a RASON JSON definition
-- **post_model_with_files** — Upload a model with file attachments (RASON + data files, or Excel)
-- **put_model** — Update an existing model or create a new version
-- **put_model_with_files** — Update a model with file attachments
-- **set_champion** — Set or unset the champion (default) version of a model
-- **delete_model** — Delete models with flexible filtering by kind, type, and force option
+- **list_models**: List models in your account with optional filters by type and kind
+- **get_model**: Retrieve a model definition (RASON JSON)
+- **post_model**: Create a new model from a RASON JSON definition
+- **post_model_with_files**: Upload a model with file attachments (RASON + data files, or Excel)
+- **put_model**: Update an existing model or create a new version
+- **put_model_with_files**: Update a model with file attachments
+- **set_champion**: Set or unset the champion (default) version of a model
+- **delete_model**: Delete models with flexible filtering by kind, type, and force option
 
 </details>
 
@@ -277,6 +288,6 @@ This software is developed and maintained by **Frontline Systems, Inc.** For que
 
 ## Links
 
-- [RASON Portal](https://rason.com) — Create a free account, manage models, documentation
-- [Frontline Systems](https://www.solver.com) — The company behind RASON and Analytic Solver
-- [MCP Protocol](https://modelcontextprotocol.io) — Learn about the Model Context Protocol
+- [RASON Portal](https://rason.com): create a free account, manage models, documentation
+- [Frontline Systems](https://www.solver.com): the company behind RASON and Analytic Solver
+- [MCP Protocol](https://modelcontextprotocol.io): learn about the Model Context Protocol

package/out/index.cjs

@@ -309,7 +309,7 @@ Data flow:`);for(let a of o)t.push(`  ${a}`)}return{lines:t,observations:r}}func
 ### Structure`),t.push(...a.lines),a.observations.length>0){t.push(`
 ### Observations`);for(let x of a.observations)t.push(`- ${x}`)}let s=new Set(["modelName","modelDescription","modelType","modelSettings","comment","variables","constraints","objective","uncertainVariables","uncertainFunctions","formulas","data","datasources","datasets","estimator","transformer","fittedModel","weakLearner","preProcessor","actions","decisionTables","boxFunctions","boxIterators","typeDefs","flowName","workflowName","flowDescription","workflowDescription","engineSettings","indexSets","parameters","evaluations","contexts"]),l=Object.keys(e).filter(x=>{if(s.has(x))return!1;let S=e[x];return!(S&&typeof S=="object"&&!Array.isArray(S)&&"modelName"in S)}),d=[],p=Ke(e.engineSettings);p&&d.push(`engineSettings (${p})`);let f=Ke(e.indexSets);f&&d.push(`indexSets (${f})`);let g=Ke(e.parameters);g&&d.push(`parameters (${g})`);let y=Ke(e.evaluations);y&&d.push(`evaluations (${y})`);let v=Ke(e.contexts);return v&&d.push(`contexts (${v})`),(d.length>0||l.length>0)&&(t.push(`
 ### Other sections`),d.length>0&&t.push(d.join(" | ")),l.length>0&&t.push(`Unrecognized: ${l.join(", ")}`)),t.join(`
-`)}function OR(e){e.registerTool("describe_model",{title:"Analyze RASON Model",description:"Analyze and classify a RASON model: reliably detect model type and subtype (optimization \u2014 LP, QP, MIP, stochastic, chance-constrained; simulation; datamining \u2014 classification, regression, clustering; calculation \u2014 decision tables, box functions; multi-stage flow), inventory and classify components (variable types, constraint types, distribution families, ML pipelines, inter-stage data flow), and identify issues (missing result requests, absent validation data, empty output sections). Provides deterministic RASON-specific analysis beyond what raw JSON inspection reveals. Call whenever the user asks about, discusses, or works with a RASON model \u2014 whether open in the editor, referenced as a file, provided inline, or stored on the RASON server.",annotations:{readOnlyHint:!0,destructiveHint:!1,idempotentHint:!0,openWorldHint:!0},inputSchema:I.object({model:I.record(I.string(),I.unknown()).optional().describe("The RASON model as a JSON object. Mutually exclusive with rason_model_path and name_or_id."),rason_model_path:I.string().optional().describe("Path to a file containing a RASON model. Mutually exclusive with model and name_or_id."),name_or_id:I.string().optional().describe("Name or ID of an existing model on the RASON account. The model will be fetched and analyzed. Mutually exclusive with model and rason_model_path.")})},async t=>{let{model:r,rason_model_path:i,name_or_id:n}=t,o=[r,i,n].filter(l=>l!==void 0).length;if(o===0)throw new Error("Must provide one of: model, rason_model_path, or name_or_id.");if(o>1)throw new Error("Provide only one of: model, rason_model_path, or name_or_id.");let a;if(r)a=r;else if(i){let l=await sn(i);try{a=JSON.parse(l)}catch(d){throw new Error(`Failed to parse JSON from ${i}: ${d instanceof SyntaxError?d.message:"unknown error"}. The file may contain syntax errors or incomplete JSON. Read the file directly to inspect and fix the content.`)}}else{let d=(await tt("GET",`/model/${encodeURIComponent(n)}`)).content[0].text;try{a=JSON.parse(d)}catch(p){throw new Error(`Failed to parse model response for "${n}": ${p instanceof SyntaxError?p.message:"unknown error"}. Use get_model to retrieve the raw content and inspect it directly.`)}}return{content:[{type:"text",text:tF(a)}]}})}var NR=process.env.RASON_EXAMPLES_DIR||AR.default.resolve(__dirname,"../examples"),Bi=new qm({name:"rason-mcp-server",version:"2026.5.3"},{instructions:`This MCP server connects to Frontline Systems' RASON REST API (endpoint: https://rason.net, web app: https://rason.com). RASON (RESTful Analytic Solver Object Notation) is a JSON-based modeling language created by Frontline Systems, Inc. (https://www.solver.com), the creator of the Excel Solver, for defining and solving analytical models via a REST API.
+`)}function OR(e){e.registerTool("describe_model",{title:"Analyze RASON Model",description:"Analyze and classify a RASON model: reliably detect model type and subtype (optimization \u2014 LP, QP, MIP, stochastic, chance-constrained; simulation; datamining \u2014 classification, regression, clustering; calculation \u2014 decision tables, box functions; multi-stage flow), inventory and classify components (variable types, constraint types, distribution families, ML pipelines, inter-stage data flow), and identify issues (missing result requests, absent validation data, empty output sections). Provides deterministic RASON-specific analysis beyond what raw JSON inspection reveals. Call whenever the user asks about, discusses, or works with a RASON model \u2014 whether open in the editor, referenced as a file, provided inline, or stored on the RASON server.",annotations:{readOnlyHint:!0,destructiveHint:!1,idempotentHint:!0,openWorldHint:!0},inputSchema:I.object({model:I.record(I.string(),I.unknown()).optional().describe("The RASON model as a JSON object. Mutually exclusive with rason_model_path and name_or_id."),rason_model_path:I.string().optional().describe("Path to a file containing a RASON model. Mutually exclusive with model and name_or_id."),name_or_id:I.string().optional().describe("Name or ID of an existing model on the RASON account. The model will be fetched and analyzed. Mutually exclusive with model and rason_model_path.")})},async t=>{let{model:r,rason_model_path:i,name_or_id:n}=t,o=[r,i,n].filter(l=>l!==void 0).length;if(o===0)throw new Error("Must provide one of: model, rason_model_path, or name_or_id.");if(o>1)throw new Error("Provide only one of: model, rason_model_path, or name_or_id.");let a;if(r)a=r;else if(i){let l=await sn(i);try{a=JSON.parse(l)}catch(d){throw new Error(`Failed to parse JSON from ${i}: ${d instanceof SyntaxError?d.message:"unknown error"}. The file may contain syntax errors or incomplete JSON. Read the file directly to inspect and fix the content.`)}}else{let d=(await tt("GET",`/model/${encodeURIComponent(n)}`)).content[0].text;try{a=JSON.parse(d)}catch(p){throw new Error(`Failed to parse model response for "${n}": ${p instanceof SyntaxError?p.message:"unknown error"}. Use get_model to retrieve the raw content and inspect it directly.`)}}return{content:[{type:"text",text:tF(a)}]}})}var NR=process.env.RASON_EXAMPLES_DIR||AR.default.resolve(__dirname,"../examples"),Bi=new qm({name:"rason-mcp-server",version:"2026.5.5"},{instructions:`This MCP server connects to Frontline Systems' RASON REST API (endpoint: https://rason.net, web app: https://rason.com). RASON (RESTful Analytic Solver Object Notation) is a JSON-based modeling language created by Frontline Systems, Inc. (https://www.solver.com), the creator of the Excel Solver, for defining and solving analytical models via a REST API.
 
 RASON model types:
 - Optimization \u2014 linear, mixed-integer, nonlinear, conic, quadratic, stochastic; powered by Gurobi, XPRESS, MOSEK, KNITRO engines

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@frontlinesystems/rason-mcp-server",
-  "version": "2026.5.3",
+  "version": "2026.5.5",
   "mcpName": "io.github.FrontlineSystems/rason-mcp-server",
   "description": "Build, solve, and analyze RASON optimization, simulation, data science, and decision models",
   "author": "Frontline Systems (https://www.solver.com)",
@@ -58,6 +58,7 @@
     "inspector:test": "npx @modelcontextprotocol/inspector node --env-file=.env.test out/index.cjs",
     "inspector:debug": "npx @modelcontextprotocol/inspector node --env-file=.env.local --inspect-brk out/index.cjs",
     "build:readme": "node scripts/build-readme.js",
+    "build:mcpb": "node scripts/build-mcpb.js",
     "prepack": "npm run build:readme && npm run compile:prod"
   },
   "dependencies": {
@@ -68,6 +69,7 @@
     "zod": "^4.3.6"
   },
   "devDependencies": {
+    "@anthropic-ai/mcpb": "^2.1.2",
     "@types/mime-types": "^3.0.1",
     "@types/node": "^22.10.5",
     "esbuild": "^0.27.4",