freelance-search

name	freelance-search
description	Search freelance platforms and score jobs against portfolio with automated extraction, LLM scoring, and proposal drafting

Freelance Job Search Skill

When to Use

• User mentions: "search jobs", "find gigs", "look for work", "freelance opportunities"
• Keywords: upwork, freelancer, guru, peopleperhour, twine
• User wants to discover and evaluate job opportunities

Quick Start

# CLI usage
freelance-ops search "python django" --min-score 70

# API usage
curl -X POST <http://localhost:8000/api/pipeline/run> \
  -H "Content-Type: application/json" \
  -d '{"query": "python django", "platforms": ["upwork", "freelancer"]}'

Workflow

1. Load Portfolio

Read portfolio from data/settings.json or prompt user to run freelance-ops init:

import json
from pathlib import Path

settings_path = Path("data/settings.json")
if settings_path.exists():
    settings = json.loads(settings_path.read_text())
    portfolio = settings.get("portfolio", {})
else:
    print("Run: freelance-ops init")
    exit(1)

2. Extract Jobs Concurrently

Use platform extractors defined in freelance_ops/extractors/:

• Upwork: API-based (upwork_extractor.py) - requires UPWORK_API_KEY
• Freelancer: Scraping with StealthyFetcher (freelancer_extractor.py)
• Guru: Scraping with session handling (guru_extractor.py)
• PeoplePerHour: Pagination support (peopleperhour_extractor.py)
• Twine: Minimal API for creative gigs (twine_extractor.py)

3. Score with LLM

Pipeline uses FreelanceScorer from freelance_ops/pipeline/scorer.py:

• 80-100: Strong match (>70% skills overlap, relevant past projects)
• 50-79: Moderate match (transferable skills, some gaps)
• 0-49: Weak match (significant skill gaps)

Scoring prompt includes:

• Portfolio skills and past projects
• Budget compatibility
• Client rating threshold
• Job description relevance

4. Present Results

Format output as Rich table (CLI) or JSON (API):

Rank | Platform | Score | Title | Missing Skills | Rationale
─────|──────────|───────|──────────|──────────────|──────────────
 1   | Upwork   | 87    | Python Developer | Docker       | Strong Django experience...
 2   | Guru     | 82    | Django Expert  | React        | Past project match...

Platform Quirks

See resources/platform-quirks.json for extractor-specific tips and rate limits.

Proposal Drafting

For jobs scoring >80, automatically draft proposals using:

• 2-3 most relevant past projects from portfolio
• Address missing skills with learning timeline
• Second-person language ("You/Your")
• 150-250 words in markdown format

Use the freelance-proposal skill for detailed proposal guidance.

Example Conversations

User: "Search for Python Django jobs on Upwork and Freelancer" Assistant:

1. Loads portfolio from settings
2. Runs extractors concurrently for both platforms
3. Scores jobs using LLM with portfolio context
4. Presents top 5 results in table format
5. Offers to draft proposals for scores >80

User: "Find gigs related to API development" Assistant:

1. Searches all 5 platforms simultaneously
2. Deduplicates by (platform, job_id)
3. Scores against portfolio
4. Filters by min_score (default 50)
5. Saves results to data/results_YYYYMMDD.json

Configuration

• Min Score: Change with --min-score flag (default: 50)
• Platforms: Modify in data/settings.json or use --platforms flag
• Max Results: Adjust with --max-results flag (default: 20 per platform)

Files Referenced

• freelance_ops/cli.py - CLI entry point
• freelance_ops/pipeline/orchestrator.py - Pipeline coordinator
• freelance_ops/pipeline/scorer.py - LLM scoring logic
• freelance_ops/extractors/*.py - Platform-specific extractors
• data/settings.json - Portfolio and schedule config