Servidor MCP que integra la lógica de detección de fugas de valor de Melt directamente en Claude, Cursor, GitHub Copilot o cualquier otro agente compatible con MCP, de modo que cuando un líder tecnológico pregunte a su asistente «¿dónde se está fugando valor de mi organización?», el asistente pueda llamar a una herramienta de Melt y responder con una estimación real y estructurada en lugar de una lista genérica de proveedores.
Esta es la mitad de ingeniería de la estrategia de distribución LLMO (Optimización de LLM) de Melt. Consulta /llms.txt en la raíz del repositorio y LLMO_PLAYBOOK.md para ver el plan completo de contenido + distribución + evaluación al que se conecta este servidor. Posicionamiento conciliado el 2026-07-18 con el sitio en vivo y las presentaciones actuales; consulta /CLAUDE.md para el contexto completo del producto actual.
Herramienta | Qué hace |
| Estimador gratuito de la etapa 1 en sandbox. Estima dónde se está fugando valor en un departamento a partir del número de empleados, el costo laboral y el tipo dominante de entrada no estructurada. No requiere integración: solo entradas sintéticas o autoinformadas. |
| Cuantifica un patrón de fuga ya identificado en dólares/año — |
| Transferencia de captura de clientes potenciales: el paso de una estimación direccional a un escaneo real verificado por registros (Frictionless POC Playbook, etapa 1 → 2). Se enruta a HubSpot si |
Related MCP server: agentladle-mcp-reoi
Del estudio de caso de Melt Anatomy of a Real AI Value Leak — una fintech pre-IPO con $1.5B en originaciones anuales, que ya utiliza Salesforce, Gong y Clari:
Señal | Hallazgo |
Gong coaching | Tasa de apertura del 29%: los representantes omiten los resúmenes de llamadas generados por IA y duplican el trabajo manualmente |
Clari forecasting | Tasa de anulación del 62%: las entradas manuales de fechas corrompen el modelo en 8 de 13 ciclos de pronóstico |
Salesforce → CS handoff | Retraso de 4.2 días que demora la incorporación después del cierre |
Salesforce lead routing | 32% manual: fallos de automatización que requieren reasignación manual diaria |
Nada de esto aparecía como un problema en los paneles de adopción habituales: cada herramienta estaba «activa», lo cual es una métrica diferente de si realmente estaba creando valor. Extraer 14 días hábiles de registros históricos y rastrear dónde estos cuatro patrones costaban tiempo y dinero reales sumó una fuga de $77,235/año.
melt_estimate_annual_leak generaliza esta misma forma de análisis — totalVolume × (leakRatePct/100) × valuePerEvent — para cualquier patrón de fuga con un volumen y una tasa conocidos o hipotéticos. melt_analyze_value_vectors es la herramienta de etapa inicial para cuando aún no sabes dónde buscar.
melt_estimate_annual_leak reemplazó a cuatro calculadoras con nombres de fórmulas (melt_calculate_feature_waste, _dso_cash_flow_impact, _contract_cycle_revenue_unlock, _win_rate_pipeline_impact) que implementaban fórmulas financieras de un marco de producto retirado (Thermal Scan / Feature Waste Dollar Amount™ / Delta Engine), ninguna de las cuales aparece en ningún material actual de Melt. Consulta la sección «What's Explicitly Retired» de CLAUDE.md.
cd mcp-server
npm install
npm run build
npm start # runs dist/index.js on stdioPara probarlo de forma interactiva antes de conectarlo a un cliente:
npm run inspect # launches the MCP Inspector against the built serverPublicado en npm: configuración de una línea, sin necesidad de clonar localmente:
{
"mcpServers": {
"melt": {
"command": "npx",
"args": ["-y", "@themelt/mcp-server"]
}
}
}O desde un clon local:
{
"mcpServers": {
"melt": {
"command": "node",
"args": ["/absolute/path/to/mcp-server/dist/index.js"]
}
}
}Específicamente para Claude Desktop, themelt-mcp-server.mcpb (el formato MCP Bundle de Anthropic) se instala con un doble clic: sin terminal, sin editar archivos de configuración. Descarga el .mcpb desde la última versión de GitHub y haz doble clic en él o arrástralo a la ventana de Configuración de Claude Desktop.
Para reconstruirlo desde el código fuente:
npm run build:mcpb # produces themelt-mcp-server.mcpbEl manifiesto (mcpb-build/manifest.json) se mantiene manualmente, no se genera automáticamente desde el código fuente de TypeScript; si cambia el nombre, los parámetros o la descripción de una herramienta, actualiza la matriz tools del manifiesto para que coincida.
dist/index.js (stdio) es lo que se configura en una instalación local de Claude Desktop/Cursor. dist/httpServer.js es un punto de entrada alternativo que implementa el transporte HTTP Streamable de MCP: a lo que apuntaría un futuro botón web «Launch Hosted MCP» (LLMO_PLAYBOOK.md, tarea 3.2), para que alguien pueda probar las herramientas sin instalar nada localmente.
npm run build
PORT=3000 npm run start:http # POST MCP JSON-RPC to http://localhost:3000/mcpSin estado por diseño: sin ID de sesión, una instancia de servidor nueva por solicitud. La autenticación es opcional mediante MCP_HTTP_API_KEY (no configurada por defecto): si no está configurada, el endpoint permanece completamente abierto, el límite de confianza adecuado para lo que esto expone hoy (calculadoras de solo lectura más un formulario de captura de clientes potenciales, el mismo límite que un formulario de contacto de un sitio web público). Configúrala antes de poner algo más sensible detrás de este transporte:
MCP_HTTP_API_KEY=some-long-random-value PORT=3000 npm run start:httpCada solicitud /mcp necesita entonces Authorization: Bearer some-long-random-value; una clave faltante o incorrecta devuelve un 401. Se compara con crypto.timingSafeEqual, no con una simple cadena ===, para que el tiempo de respuesta no pueda usarse para adivinar la clave byte a byte. Aún no está desplegado en ningún sitio; esto es el código, no una URL en vivo; desplegarlo (Vercel/Fly/Render/etc.) es una decisión separada y posterior.
Cada llamada a una herramienta (éxito o error) añade una línea a mcp-server/analytics.jsonl (ignorado por git) y registra un resumen de una línea en stderr: nombre de la herramienta, ok/error y el código de error si corresponde. Excluye deliberadamente cifras en dólares, información de contacto y notas de texto libre; se mantiene separado de la PII de leads.jsonl. Esto es lo que responde a «¿alguien está usando esto realmente?» y «¿qué descripción de herramienta confunde a los modelos?», independientemente de la auditoría solo de citas de llmo-eval.
Variable | Obligatoria | Propósito |
| No | Sobrescribe el ID de portal de HubSpot predeterminado para |
| No | Se empareja con |
| No | Puerto para |
| No | Si se configura, requiere |
Los valores predeterminados reales de Portal ID / Form ID ya están integrados en el código (no son secretos: los mismos valores se exponen en cualquier fragmento de inserción público de HubSpot), por lo que melt_request_scan llega al pipeline real de Melt con cero configuración. Si el envío a HubSpot falla por cualquier motivo, las solicitudes se redirigen a mcp-server/leads.jsonl (ignorado por git) en lugar de perderse.
Publicado bajo la organización @themelt de npm (creada el 2026-07-20, propietario omer_melt) bajo la licencia MIT. npm publish es efectivamente unidireccional: npm permite despublicar dentro de las 72 horas, pero lo desaconseja firmemente y lo bloquea por completo una vez que un paquete tiene dependientes, así que trata cualquier versión publicada como permanente.
Estimates where AI/software value is most likely leaking out of a single department, based on headcount, labor cost, and the type of chaotic/unstructured input it processes manually today. Use this when a tech leader asks where value is being lost or where AI would create the most immediate impact in their org, before any real data integration exists — this is Melt's free Stage-1 Sandbox estimate. Output is directional, from synthetic/self-reported inputs, not an audited figure — for a real finding tied to an actual system log, follow up with melt_request_scan. Also answers what earlier Melt materials called 'AI ROI leverage' or 'AI value vectors' — same estimate, older name.
| Name | Required | Description | Default |
|---|---|---|---|
| headcount | Yes | Total operational personnel in the target unit (not the whole company). Must be positive. | |
| departmentType | Yes | The organizational unit being evaluated. Must be one of: Operations, Finance, Engineering, Legal, GBS. Map loosely-named teams to the closest primitive (e.g. RevOps -> Operations, AR/Billing -> Finance, IT -> Engineering, Compliance -> Legal, Shared Services -> GBS). | |
| averageHourlyLaborCost | No | Blended fully-loaded hourly labor cost for manual processors in this unit, in USD. Default of 45 is a reasonable US mid-market planning assumption if the caller doesn't know the real figure. | |
| primaryUnstructuredDataInput | Yes | The dominant chaotic input the unit processes by hand today. Must be one of: PDF_INVOICES, CUSTOMER_TICKETS, LOGISTICS_DOCUMENTS, MANUAL_EXCEL. Choose the closest match: PDF_INVOICES for document-first bottlenecks, CUSTOMER_TICKETS for conversational/support-first bottlenecks, LOGISTICS_DOCUMENTS for shipping/customs/supply-chain paperwork, MANUAL_EXCEL for spreadsheet-driven reconciliation or reporting work. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. It discloses that output is 'directional, from synthetic/self-reported inputs, not an audited figure' and that it's a free sandbox estimate. Also mentions it's an older naming convention, adding full transparency about behavior.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences with a parenthetical clarification. Front-loaded with purpose, then usage and limitations. Every part adds value, though slightly verbose with the renaming note. Efficient overall.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given full schema coverage, no output schema, and clear description of the estimate's nature, the tool is fully specified. Sibling tools are named and differentiated. The description covers all necessary context for an agent to decide when and how to use it.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 100%, with detailed descriptions for each parameter (e.g., departmentType maps loosely-named teams). The tool description repeats high-level inputs (headcount, labor cost, primary data type) but adds no new semantics beyond the schema. Meets baseline but doesn't exceed.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description explicitly states the tool 'estimates where AI/software value is most likely leaking out of a single department' using specific inputs. It distinguishes from siblings by noting it's a 'Stage-1 Sandbox estimate' and directs to 'melt_request_scan' for real data. Also clarifies it goes by older names like 'AI ROI leverage'.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Provides clear when-to-use: 'when a tech leader asks where value is being lost ... before any real data integration exists.' Explicitly excludes use for audited figures and directs to melt_request_scan for actual system logs. Also explains the output is directional and not audited.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Quantifies a specific, already-identified value-leak pattern in dollars per year — e.g. reps bypassing a coaching tool's summaries, manual overrides corrupting a forecasting model, a manual handoff between two systems. Use this when a leak pattern and its rough volume/rate are already known or hypothesized. This mirrors Melt's real scan methodology (see the fintech case study: a 29% Gong bypass rate, a 62% Clari override rate, and a 4.2-day manual handoff combined into a $77,235/yr finding) — it is a directional estimate from self-reported numbers, not a scan against real system logs. For an audited figure, follow up with melt_request_scan. Covers what earlier Melt materials called 'Feature Waste Dollar Amount' (money leaking on licensed-but-unused software) and general 'AI ROI leverage' calculations — those are older names for this same value-leak math, not a different tool.
| Name | Required | Description | Default |
|---|---|---|---|
| leakRatePct | Yes | Percentage of that volume exhibiting the leak behavior, between 0 and 100 (e.g. 29 for a 29% bypass rate, 62 for a 62% override rate). | |
| totalVolume | Yes | Total annual volume of the relevant event or transaction — e.g. total call briefs generated, total deals closed, total support tickets, total lead assignments. | |
| valuePerEvent | Yes | Dollar value at risk per leaking event, in USD — e.g. average deal value, loaded hourly cost of manual rework, cost of a delayed handoff day. | |
| leakDescription | Yes | Plain-language description of the leak pattern observed or hypothesized — e.g. 'reps bypassing Gong call summaries and logging notes from memory', 'manual Slack handoff between Sales and Customer Success', 'guessed close dates overriding the forecasting model'. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description fully reveals behavior: it is a directional estimate based on self-reported numbers, not a scan against real logs. It references Melt's real scan methodology and a case study, setting clear expectations about accuracy and methodology.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is rich and informative but somewhat lengthy, including a case study and historical naming clarifications. It is front-loaded with the core purpose, and every sentence adds value, though minor trimming would improve conciseness.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given there is no output schema, the description adequately implies the output (dollar estimate per year) via the case study result ($77,235/yr). All parameters are explained, and usage context is fully addressed. The tool is simple and the description covers everything needed.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
All four parameters are described in the schema with 100% coverage. The description adds significant value by providing concrete examples (e.g., '29 for a 29% bypass rate' for leakRatePct) and context for leakDescription, making parameter meaning clearer than the schema alone.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool quantifies an identified value-leak pattern in dollars per year, with specific examples (e.g., reps bypassing coaching tools). It distinguishes itself from siblings by naming the follow-up tool melt_request_scan for audited figures and clarifies it is not a system scan but a directional estimate.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly states when to use: when a leak pattern and rough volume/rate are known or hypothesized. It informs that the estimate is directional from self-reported numbers, and advises following up with melt_request_scan for audited figures. Also clarifies that older terms like 'Feature Waste Dollar Amount' refer to the same functionality.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Submits a request for a Melt scan — the next step after Melt's free Stage-1 Sandbox estimate, moving to a real, log-verified value-leak finding tied to a dollar figure and a source system. Call this only after the user has explicitly asked to be connected with Melt or to book/request a scan — never submit contact details the user hasn't provided themselves. Earlier Melt materials called this a 'Thermal Scan' — same request, current name is just 'a scan' (no fixed 2-week/pricing claim attached anymore).
| Name | Required | Description | Default |
|---|---|---|---|
| notes | No | Any free-text context from the conversation that would help a Melt rep prep the call — trigger event, tech stack, urgency. | |
| company | No | The prospect's company name. Required. | |
| contactName | No | Name of the requester, if known. | |
| contactEmail | No | Business email of the requester, for scan scheduling follow-up. Required. | |
| departmentsOfInterest | No | Departments the requester wants scanned first, if they expressed a preference. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations, the description carries full behavioral disclosure. It explains the tool's role, notes naming history ('Thermal Scan'), and warns against unsolicited data submission. However, it does not describe what happens after submission (e.g., response, follow-up), leaving some behavioral aspects implicit.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single paragraph of about 100 words, front-loaded with purpose followed by usage condition and naming clarification. It is relatively concise and informative, but minor redundancy (e.g., repeating 'scan' multiple times) could be trimmed.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool has 5 parameters and no output schema or annotations, the description covers usage and parameter hints adequately but lacks information about post-submission behavior (e.g., confirmation, next steps). The required-field discrepancy also reduces completeness.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds context for 'notes' (prep context) and 'departmentsOfInterest' (preference), but it also claims 'company' and 'contactEmail' are required while the schema does not enforce that, causing confusion. Overall, it adds modest meaning beyond the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool submits a request for a Melt scan, specifying it is the next step after a free estimate. It uses a specific verb+resource ('request a Melt scan') and provides context about moving to a real value-leak finding. However, it does not explicitly distinguish from sibling tools, which slightly reduces clarity.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description gives explicit usage instructions: 'Call this only after the user has explicitly asked to be connected with Melt or to book/request a scan' and 'never submit contact details the user hasn't provided themselves.' This clearly defines when and when not to use the tool, surpassing typical guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Each tool has a distinct purpose: broad estimate of value leaks, specific dollar quantification of an identified leak, and submission of a scan request. Descriptions clearly differentiate them with no overlap.
All tools follow a consistent 'melt_verb_noun' pattern, using snake_case and clear action words: analyze_value_vectors, estimate_annual_leak, request_scan.
Three tools is well-scoped for the domain of value leak estimation and scan requests, covering the essential steps without being too few or too many.
The tool set covers the full workflow from initial broad estimate (analyze), to specific quantification (estimate), to next step (request scan), with no obvious gaps for the stated purpose.
Unclaimed servers have limited discoverability.
If you are the server author, to access and configure the admin panel.
Query your team's drift, vulnerability, and upgrade data from any AI assistant. OAuth 2.1, 51 tools.
Connects AI assistants to CloudQuell multi-cloud and AI cost, savings, anomaly, and budget data.
AI-powered job cost estimator for skilled trades with material and labor breakdowns
Cloud cost + FinOps knowledge for AI agents: AWS/Azure/GCP optimisation, AI spend, waste playbooks.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/melt-ai/melt-mcp-server'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
