Existing issue
Product variant
Cloud
Area
AI & Agent
Problem
Reactive Resume already provides MCP support, but the MCP experience should evolve toward full feature parity with the platform itself.
An AI agent connected through MCP should be able to perform the same meaningful resume-management operations that a user can perform through the Reactive Resume interface, without requiring the user to manually open the platform for each step.
Today, an agent can interact with Reactive Resume through MCP, but the experience should be comprehensive enough to support the full resume creation and editing workflow, including both content and visual presentation.
A particularly important limitation is that generating or modifying resume data is not enough by itself. A resume can contain correct information while still having poor pagination, spacing, typography, section ordering, density, or visual hierarchy.
For an agent-driven workflow to be genuinely useful, the agent should be able to create and modify the resume and also control its visual configuration, verify the resulting layout, and iterate until the final document is properly formatted and ready to export.
The goal is for Reactive Resume to become a first-class tool for AI agents, rather than requiring agents to manipulate raw resume data and then rely on the user to manually fix the visual result.
Desired outcome
Expand the existing Reactive Resume MCP so that an MCP-compatible agent can perform the complete resume workflow through the same underlying capabilities available to users in the platform.
The MCP should support both content and presentation.
An agent should be able to:
- List and inspect resumes.
- Create resumes.
- Read resume contents.
- Update resume contents.
- Delete resumes.
- Duplicate resumes.
- Import and export supported resume formats.
- Generate and retrieve exported documents such as PDF.
- Work with stored reusable sources.
- Create a resume from an existing source.
- Selectively import information from a source.
- Manage sections, entries, ordering, and other resume content.
- Access and modify resume templates.
- Change fonts.
- Change colors.
- Change spacing and typography.
- Change page and layout settings.
- Reorder sections.
- Adjust visual hierarchy and formatting.
- Configure the resume for different presentation goals.
The MCP should also allow an agent to reason about the final rendered resume rather than only manipulating its underlying JSON.
For example, an agent should be able to receive a request such as:
"Create a Java Backend resume using my information from Reactive Resume. Use my relevant experience, education, projects and skills, select an appropriate professional layout, optimize the typography and spacing, and make sure the final resume fits correctly before exporting it."
The agent should then be able to:
- Query the user's available information.
- Select relevant content.
- Create or modify the resume.
- Configure its visual presentation.
- Render or preview the result.
- Detect layout problems such as excessive whitespace, awkward page breaks, overflow, inconsistent spacing, or poor information density.
- Make the necessary adjustments.
- Verify the final result.
- Export the completed resume.
The objective is for an agent to be able to take a resume from raw information to a polished, visually coherent, export-ready document without requiring manual cleanup in the web interface.
Alternatives considered
The current alternative is to use the Reactive Resume web interface manually after an agent has generated or modified the resume data.
Another option is for an agent to generate a Resume JSON file and have the user import it into Reactive Resume. This still leaves the user responsible for choosing the template, adjusting typography, fixing layout issues, checking pagination, and preparing the final document.
A separate third-party MCP wrapper around the Reactive Resume API can also be used. However, this does not provide the same level of integration as having the MCP capabilities maintained as a first-class part of Reactive Resume itself.
The existing Reactive Resume MCP should therefore be extended rather than requiring users to rely on external wrappers or manually switch between an AI agent and the web interface.
Proposed scope
The scope should extend the existing MCP rather than introduce a separate, incompatible MCP implementation.
Included:
- Bring MCP capabilities toward feature parity with the core Reactive Resume resume-management functionality.
- Support authenticated access using the user's existing Reactive Resume API key.
- Support standard MCP clients such as OpenCode, Claude Code, Gemini CLI, Codex, and other MCP-compatible clients.
- Provide read and write operations for resumes.
- Support complete resume lifecycle operations.
- Support importing and exporting supported resume formats.
- Support reusable resume sources when available.
- Allow agents to create resumes from existing sources and selectively use their content.
- Expose resume templates and relevant configuration options.
- Allow agents to modify visual configuration, including fonts, colors, spacing, layout, section ordering, and other supported presentation settings.
- Provide a way for agents to render or preview a resume after making changes.
- Make it possible for agents to verify the rendered result before considering the resume complete.
- Allow agents to iteratively correct visual/layout problems.
- Support final export through MCP.
The visual verification workflow should ideally account for practical document-quality issues such as:
- Unexpected page breaks.
- Content overflow.
- Excessive empty space.
- Inconsistent spacing.
- Poor typography.
- Unbalanced page density.
- Sections being split awkwardly.
- Important information being pushed to an undesirable page.
- Layout problems introduced by changing fonts or content length.
Out of scope for the initial implementation:
- Supporting non-MCP AI agent protocols.
- Requiring a specific AI provider or model.
- Tying the MCP to a specific client such as OpenCode, Claude Code, Gemini CLI, or Codex.
- Replacing the existing web editor.
- Automatically deciding what information is truthful or inventing professional experience.
- External sources such as GitHub or Google Drive unless they are already supported independently by Reactive Resume.
The implementation should reuse the existing Reactive Resume APIs, schemas, rendering pipeline, templates, and business logic wherever possible rather than creating a second independent representation of resumes.
Additional context
Reactive Resume already has an MCP implementation, API-key authentication, MCP tooling, and a dedicated resume-builder skill. The purpose of this proposal is therefore to evolve the existing integration into a complete agent interface rather than introduce MCP from scratch.
Reactive Resume's current architecture already separates MCP tooling, resume schemas, PDF rendering, DOCX export, and import functionality, which should make it possible to expose more of these existing capabilities through MCP without duplicating the underlying resume logic.
A key use case is an agent-driven end-to-end workflow:
User
↓
OpenCode / Claude Code / Gemini CLI / Codex
↓
Reactive Resume MCP
↓
Read user's professional information
↓
Select relevant content
↓
Create or update resume
↓
Configure template and visual design
↓
Render / preview
↓
Check layout and pagination
↓
Iterate if necessary
↓
Export final resume
For example:
"Create a Java Backend resume for this job description using my information from Reactive Resume. Include my relevant experience, education and projects. Use an appropriate professional template, optimize the typography and spacing, make sure it fits correctly, and export the final PDF."
The agent should be able to complete this workflow through MCP without requiring the user to manually open Reactive Resume and correct the result.
The visual aspect is especially important. An MCP that only modifies structured resume data can still produce a technically valid but visually poor resume. The agent should therefore have enough control and feedback from the rendering layer to make informed visual adjustments.
This could also enable reusable agent skills on top of Reactive Resume, such as:
- Tailoring a resume to a job description.
- Creating role-specific resumes from a master profile.
- Optimizing a resume's visual hierarchy.
- Preparing ATS-friendly versions.
- Translating a resume while preserving its layout.
- Creating multiple resume variants.
- Reviewing an existing resume and automatically applying formatting improvements.
The MCP should remain a standard, client-agnostic MCP interface so that these workflows can work across OpenCode, Claude Code, Gemini CLI, Codex, and other compatible clients.
The existing MCP should be treated as the foundation; this proposal is about expanding its capabilities and making it capable of producing complete, polished, export-ready resumes.
Existing issue
Product variant
Cloud
Area
AI & Agent
Problem
Reactive Resume already provides MCP support, but the MCP experience should evolve toward full feature parity with the platform itself.
An AI agent connected through MCP should be able to perform the same meaningful resume-management operations that a user can perform through the Reactive Resume interface, without requiring the user to manually open the platform for each step.
Today, an agent can interact with Reactive Resume through MCP, but the experience should be comprehensive enough to support the full resume creation and editing workflow, including both content and visual presentation.
A particularly important limitation is that generating or modifying resume data is not enough by itself. A resume can contain correct information while still having poor pagination, spacing, typography, section ordering, density, or visual hierarchy.
For an agent-driven workflow to be genuinely useful, the agent should be able to create and modify the resume and also control its visual configuration, verify the resulting layout, and iterate until the final document is properly formatted and ready to export.
The goal is for Reactive Resume to become a first-class tool for AI agents, rather than requiring agents to manipulate raw resume data and then rely on the user to manually fix the visual result.
Desired outcome
Expand the existing Reactive Resume MCP so that an MCP-compatible agent can perform the complete resume workflow through the same underlying capabilities available to users in the platform.
The MCP should support both content and presentation.
An agent should be able to:
The MCP should also allow an agent to reason about the final rendered resume rather than only manipulating its underlying JSON.
For example, an agent should be able to receive a request such as:
"Create a Java Backend resume using my information from Reactive Resume. Use my relevant experience, education, projects and skills, select an appropriate professional layout, optimize the typography and spacing, and make sure the final resume fits correctly before exporting it."
The agent should then be able to:
The objective is for an agent to be able to take a resume from raw information to a polished, visually coherent, export-ready document without requiring manual cleanup in the web interface.
Alternatives considered
The current alternative is to use the Reactive Resume web interface manually after an agent has generated or modified the resume data.
Another option is for an agent to generate a Resume JSON file and have the user import it into Reactive Resume. This still leaves the user responsible for choosing the template, adjusting typography, fixing layout issues, checking pagination, and preparing the final document.
A separate third-party MCP wrapper around the Reactive Resume API can also be used. However, this does not provide the same level of integration as having the MCP capabilities maintained as a first-class part of Reactive Resume itself.
The existing Reactive Resume MCP should therefore be extended rather than requiring users to rely on external wrappers or manually switch between an AI agent and the web interface.
Proposed scope
The scope should extend the existing MCP rather than introduce a separate, incompatible MCP implementation.
Included:
The visual verification workflow should ideally account for practical document-quality issues such as:
Out of scope for the initial implementation:
The implementation should reuse the existing Reactive Resume APIs, schemas, rendering pipeline, templates, and business logic wherever possible rather than creating a second independent representation of resumes.
Additional context
Reactive Resume already has an MCP implementation, API-key authentication, MCP tooling, and a dedicated resume-builder skill. The purpose of this proposal is therefore to evolve the existing integration into a complete agent interface rather than introduce MCP from scratch.
Reactive Resume's current architecture already separates MCP tooling, resume schemas, PDF rendering, DOCX export, and import functionality, which should make it possible to expose more of these existing capabilities through MCP without duplicating the underlying resume logic.
A key use case is an agent-driven end-to-end workflow:
User
↓
OpenCode / Claude Code / Gemini CLI / Codex
↓
Reactive Resume MCP
↓
Read user's professional information
↓
Select relevant content
↓
Create or update resume
↓
Configure template and visual design
↓
Render / preview
↓
Check layout and pagination
↓
Iterate if necessary
↓
Export final resume
For example:
"Create a Java Backend resume for this job description using my information from Reactive Resume. Include my relevant experience, education and projects. Use an appropriate professional template, optimize the typography and spacing, make sure it fits correctly, and export the final PDF."
The agent should be able to complete this workflow through MCP without requiring the user to manually open Reactive Resume and correct the result.
The visual aspect is especially important. An MCP that only modifies structured resume data can still produce a technically valid but visually poor resume. The agent should therefore have enough control and feedback from the rendering layer to make informed visual adjustments.
This could also enable reusable agent skills on top of Reactive Resume, such as:
The MCP should remain a standard, client-agnostic MCP interface so that these workflows can work across OpenCode, Claude Code, Gemini CLI, Codex, and other compatible clients.
The existing MCP should be treated as the foundation; this proposal is about expanding its capabilities and making it capable of producing complete, polished, export-ready resumes.