Servidor del Protocolo de Contexto de Modelo (MCP) que analiza un CV frente a 5 parsers de ATS reales.
Expone tres herramientas MCP para que cualquier agente de IA compatible con MCP (Claude Code, Cursor, Windsurf, Zed, OpenCode, Cline, etc.) pueda analizar un CV frente a las peculiaridades documentadas de:
Workday
Greenhouse
Lever
Taleo (Oracle)
iCIMS
Estos son los 5 sistemas ATS empresariales más utilizados. Cada uno analiza los CV de forma ligeramente distinta. Una "puntuación ATS de 0 a 100" promedia esas diferencias convirtiéndolas en ruido. Este servidor muestra lo que realmente hace cada uno (manejo de columnas, eliminación de emojis, descarte de encabezados y pies de página, peculiaridades de formato de fecha) como hallazgos de linting concretos con soluciones concretas.
Las reglas de linting se derivan de la documentación pública de los proveedores. Consulta docs/vendor-sources.md para ver las citas.
Related MCP server: decroche-mcp
Herramienta | Descripción |
| Informe completo. Toma una ruta de archivo de CV (PDF o DOCX) y devuelve hallazgos de linting por proveedor con gravedad ( |
| Linting de un solo proveedor. Toma una ruta + nombre del proveedor. Más rápido cuando al usuario solo le interesa un ATS. |
| Devuelve la URL del compañero web visual (cv-mirror-web.vercel.app) para usuarios que prefieren una vista de parser comparativa en su navegador en lugar de una sesión de agente. |
npm install -g cv-mirror-mcpAñade esto al archivo de configuración de tu cliente MCP:
Claude Code (~/.claude/mcp.json o .claude.json en el proyecto):
{
"mcpServers": {
"cv-mirror": {
"command": "npx",
"args": ["-y", "cv-mirror-mcp"]
}
}
}Cursor / Windsurf / Zed / Cline: mismo formato de configuración (pega donde tu cliente lea mcpServers).
Reinicia el cliente. Las tres herramientas anteriores se autorregistran.
Una vez instalado, simplemente pregúntale a tu agente:
Escanea mi CV en
~/Documents/resume.pdfy dime qué haría cada ATS con él.
El agente llama a analyze_cv, el motor de linting se ejecuta localmente (sin red, sin subidas) y el agente muestra la salida estructurada:
CV Mirror — multi-vendor ATS lint report
Source: /Users/jane/Documents/resume.pdf
Format: pdf
Pages: 2
Words: 542
[ERROR] Workday
- ERROR WORKDAY_MULTI_COLUMN: 35% of lines look multi-column. Workday's parser
reads left-to-right and interleaves both columns into one stream.
Fix: Convert to single-column layout. Move sidebars (Skills, Tools, Languages)
above or below the main content.
[OK] Greenhouse
No issues detected by the simulated parser.
[WARN] Lever
- WARN LEVER_HEADER_FOOTER: Header/footer-like text detected ("Page 1 of 2").
Lever historically drops content placed in PDF headers/footers.
Fix: Remove headers and footers. Page numbers are not needed on a CV.
...Estas son heurísticas reales derivadas de documentos públicos de ATS y artículos de soporte de proveedores. La lista completa con citas se encuentra en docs/vendor-sources.md.
Regla | Proveedor | Gravedad | Disparador |
| Workday | error | >15% de las líneas tienen un espacio de 5+ caracteres |
| Workday | warn | Las fechas usan el formato "Q3 2024" |
| Workday | error | Sin correo electrónico o teléfono en texto plano |
| Greenhouse | warn | Cualquier punto de código emoji detectado |
| Greenhouse | info | Glifos de viñetas no estándar |
| Greenhouse | warn | "Mi historia", "Lo más destacado", etc. |
| Lever | warn | Patrón "Página X de Y" detectado |
| Lever | error | Sin secciones de "Experiencia"/"Educación" |
| Taleo | warn | Las fechas ISO superan a las fechas Mes-Año |
| Taleo | error | <1 palabra por kB de tamaño de archivo |
| Taleo | info | Comillas tipográficas detectadas |
| iCIMS | error | >20% de las líneas con varias columnas |
| iCIMS | warn | <2 encabezados estándar encontrados |
Node.js >= 18
Un cliente compatible con MCP (Claude Code, Cursor, Windsurf, Zed, Cline, etc.)
El servidor se ejecuta localmente dentro del proceso de tu agente. No hay punto final de carga. No hay telemetría. Los bytes de tu CV nunca salen de tu máquina.
El compañero web visual en cv-mirror-web.vercel.app también es totalmente del lado del cliente: mismo motor, se ejecuta en el navegador.
git clone https://github.com/goofypluto999/cv-mirror-mcp.git
cd cv-mirror-mcp
npm install
npm test19 pruebas unitarias que cubren la extracción de señales y la activación de reglas por proveedor.
CV Mirror (web) — vista de parser visual comparativa. Mismo motor de linting. Arrastra y suelta un PDF, mira la superposición del orden de lectura en el documento real.
Vantage AI — una vez que tu CV pasa los parsers, el siguiente problema es la solicitud: carta de presentación personalizada, entrevista simulada, análisis de ajuste. Vantage maneja ese flujo. De pago (£5 inicial / 20 tokens, 10 tokens gratis al registrarse).
Las solicitudes de extracción (pull requests) son bienvenidas. Especialmente interesado en:
Más simuladores de proveedores (BambooHR, SmartRecruiters, JazzHR, Recruitee, etc.)
Reglas actualizadas cuando los proveedores cambian su comportamiento de análisis (abre un issue con el enlace a la fuente)
Traducciones de la salida del linting
MIT. Workday, Greenhouse, Lever, Taleo e iCIMS son marcas comerciales de sus respectivos propietarios. Este proyecto no está afiliado a ninguno de ellos; los nombres se utilizan como referencia descriptiva (uso legítimo nominativo).
Creado por Vantage Labs.
Analyse a CV (PDF or DOCX) against 5 real ATS parsers (Workday, Greenhouse, Lever, Taleo, iCIMS). Returns per-vendor lint findings, parse risk score, and concrete fixes. Use when the user asks 'is my CV ATS-friendly', 'will my resume pass [vendor]', or 'why am I not getting interviews' (with a file path).
| Name | Required | Description | Default |
|---|---|---|---|
| path | Yes | Absolute path to the CV file (PDF or DOCX). |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations, so description carries full burden. It discloses tool reads PDF/DOCX, runs against 5 parsers, and returns findings. Does not mention file size limits, processing duration, or if file is uploaded elsewhere, but is largely transparent about its operation.
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 focused sentences: first defines action and output, second provides usage examples. No unnecessary words. Excellent front-loading of purpose.
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?
Despite no output schema, description explains return types (per-vendor lint findings, risk score, fixes). Mentions supported file types. Could add error handling details (e.g., missing file), but otherwise complete for a single-parameter tool.
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?
Input schema has 100% coverage with description for the single 'path' parameter. The description does not add further details beyond schema, but schema itself is sufficient. Baseline 3 applies.
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 analyzes CVs against 5 ATS parsers, returning per-vendor lint findings, risk score, and fixes. It clearly distinguishes from siblings by covering multiple vendors (vs. lint_for_vendor which likely targets one).
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?
Description provides explicit user query triggers ('is my CV ATS-friendly', 'will my resume pass [vendor]', 'why am I not getting interviews') and mentions file path requirement. Lacks explicit when-not-to-use or mention of sibling alternatives, but context is clear.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Returns the URL for the visual web companion tool (CV Mirror) where the user can drag-drop a CV and see the side-by-side parser view in their browser. Useful when the user wants the visual reading-order overlay or doesn't want to share a file path.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations provided, so description carries full burden. States it returns a URL, implying read-only, but does not explicitly declare non-destructive behavior or other constraints.
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, no unnecessary words, front-loaded with purpose. Highly efficient.
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?
Adequate for a zero-parameter, no-output-schema tool. Explains function and usage context. Minor gap: doesn't explicitly state no input needed, but schema implies 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?
No parameters (schema coverage 100%), baseline 4. Description adds meaning by explaining the purpose of the URL beyond the empty 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?
Clear verb+resource: 'Returns the URL' for a specific visual tool (CV Mirror). Distinguishes from siblings (analyze_cv, lint_for_vendor) by offering a different capability.
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?
States when it's useful (visual reading-order overlay, avoiding file path sharing). Does not explicitly exclude alternative uses but context is clear.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Run lint for ONE specific ATS vendor only. Use when the user asks something vendor-specific like 'will my CV pass Workday' or 'what would Greenhouse strip from this'. Vendor must be one of: workday, greenhouse, lever, taleo, icims.
| Name | Required | Description | Default |
|---|---|---|---|
| path | Yes | Absolute path to the CV file (PDF or DOCX). | |
| vendor | Yes | ATS vendor to simulate. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries full burden. It does not explain whether the tool is read-only, what side effects exist, or what the output format is. The term 'lint' implies analysis but lacks detail.
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, front-loaded with the core action and constraint. Each sentence adds value: one states the purpose and allowed vendors, the other gives usage examples. No wasted words.
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?
Despite its simplicity, the tool has no output schema and the description does not explain what the lint result looks like (e.g., a score, a list of issues). The user cannot infer the return format without additional context.
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?
Input schema has 100% coverage with clear descriptions. The description adds context for the vendor enum by specifying use cases, but adds no extra meaning for the path parameter beyond its schema description.
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 states 'Run lint for ONE specific ATS vendor only', with a clear verb and resource. It lists the allowed vendors and uses examples to distinguish from siblings like analyze_cv, making the purpose unambiguous.
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 provides when-to-use examples ('when the user asks something vendor-specific like...'). Does not explicitly state when not to use or name an alternative tool, though the sibling names imply a general CV analysis tool.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.0analyze_cvget_express_urllint_for_vendorScored across 3 tools
The tools have distinct purposes: analyze_cv for all vendors, lint_for_vendor for a single vendor, and get_express_url for a visual URL. However, analyze_cv and lint_for_vendor both analyze CVs, which could cause minor confusion if descriptions are not read carefully.
Tool names follow a verb_noun pattern but with inconsistency: 'analyze_cv' and 'get_express_url' are direct, while 'lint_for_vendor' uses a preposition. The verb 'lint' is less standard than 'analyze'.
With 3 tools, the set is small but well-scoped for the domain of CV ATS analysis. It covers the essential operations without being overly sparse.
The tool set covers comprehensive analysis, vendor-specific linting, and a visual companion tool. Minor gaps like listing vendors or handling multiple files are absent but not critical for the core functionality.
Generate tailored, ATS-optimized resume PDFs and cover letters from a job description, over MCP.
Public MCP server for discovering open jobs. Search, filter, and get application links.
Document-to-Markdown MCP server — convert PDF, Office and HTML into LLM-ready Markdown.
Document conversion MCP server: PDF to Markdown, image OCR, spreadsheet parsing.
