upwork-mcp

Upwork MCP Server

mcp-name: io.github.wmarceau/upwork

A Model Context Protocol (MCP) server for Upwork job discovery, client analysis, and proposal assistance. Helps freelancers find relevant opportunities and craft winning proposals.

Features

• Job Search - Search Upwork marketplace with keyword, budget, and client quality filters
• Job Details - Get full job descriptions, requirements, and screening questions
• Client Analysis - Analyze client reliability, hire rate, and spending history
• Proposal Drafting - Generate customized proposal drafts based on job requirements
• Profile Management - View your profile, stats, and portfolio
• Proposal Tracking - Track submitted proposals and their status
• Contract History - View active and past contracts with earnings

Installation

pip install upwork-mcp

Setup

1. Get Upwork API Credentials

• Go to Upwork Developer Portal
• Create a new API application
• Note your client_id and client_secret

2. Configure Environment Variables

export UPWORK_CLIENT_ID="your_client_id"
export UPWORK_CLIENT_SECRET="your_client_secret"
export UPWORK_REDIRECT_URI="https://localhost"  # or your callback URL

3. Add to Claude Desktop

Add to your Claude Desktop configuration (~/.claude/claude_desktop_config.json):

{
  "mcpServers": {
    "upwork": {
      "command": "upwork-mcp",
      "env": {
        "UPWORK_CLIENT_ID": "your_client_id",
        "UPWORK_CLIENT_SECRET": "your_client_secret"
      }
    }
  }
}

4. Authenticate

On first use, run the upwork_auth_status tool to get the authorization URL. Visit the URL, authorize the app, and use upwork_complete_auth with the callback URL.

Available Tools

Job Discovery

upwork_search_jobs

Search for jobs matching your skills and preferences.

Input:
- query: Search keywords (required)
- min_budget: Minimum budget filter
- max_budget: Maximum budget filter
- payment_verified: Only verified clients (default: true)
- limit: Max results (default: 20)

Output: Job listings with title, budget, client info, skills

upwork_get_job_details

Get full details about a specific job.

Input:
- job_id: Upwork job ID (required)

Output: Full description, questions, attachments, client details

Client Intelligence

upwork_analyze_client

Analyze a client's reliability and hiring patterns.

Input:
- job_id: Job ID to analyze client from (required)

Output: Hire rate, spend history, red flags, recommendation

Proposal Assistance

upwork_draft_proposal

Generate a customized proposal draft.

Input:
- job_id: Job ID to write proposal for (required)
- highlight_skills: Skills to emphasize
- relevant_experience: Experience to mention
- proposed_rate: Your rate

Output: Cover letter template, question answers, rate

Note: Proposals must be submitted manually. Automated submission is not allowed by Upwork.

Profile & Tracking

upwork_get_my_profile

View your freelancer profile and stats.

upwork_get_my_proposals

Track your submitted proposals.

Input:
- status: PENDING, INTERVIEWING, ARCHIVED, or ALL
- limit: Max results

upwork_get_my_contracts

View your contracts and earnings.

Input:
- status: ACTIVE, ENDED, or ALL
- limit: Max results

Authentication

upwork_auth_status

Check authentication status and get auth URL if needed.

upwork_complete_auth

Complete OAuth flow with callback URL.

Example Usage

Finding Jobs

User: Search for Amazon FBA analytics jobs with budget over $500

Claude: [Uses upwork_search_jobs with query="amazon fba analytics" min_budget=500]

Evaluating Opportunities

User: Analyze the client for job 123456

Claude: [Uses upwork_analyze_client]

Result shows:
- Client has posted 45 jobs, hired 38 (84% hire rate)
- Total spend: $125,000
- Payment verified
- Rating: LOW RISK ✅

Drafting Proposals

User: Draft a proposal for job 123456, emphasizing my Python and SP-API experience

Claude: [Uses upwork_draft_proposal with highlight_skills and experience]

Rate Limits

• Upwork API: 300 requests/minute per IP
• GraphQL queries are batched where possible

Privacy & Security

• OAuth tokens are stored locally at ~/.upwork/token.json
• No data is sent to third parties
• API requests go directly to Upwork

License

MIT License - see LICENSE

Support

• Issues: GitHub Issues
• Upwork API Docs: developer.upwork.com