The Universal Marketplace for AI Agents

Discover, evaluate, and integrate RAGs, datasets, MCP servers, APIs, documents, and agent components. Everything your AI agent needs to complete any task - all in one intelligent marketplace.

Axiomeer is a production-ready AI Agent Marketplace that serves as the central hub where AI agents discover and access everything they need:

• RAG Systems - Pre-built retrieval augmented generation pipelines
• Datasets - Curated data collections for training and inference
• MCP Servers - Model Context Protocol integrations
• APIs - 91+ external service integrations
• Documents - Pre-processed knowledge bases and documentation
• Agent Components - Reusable agent building blocks
• Tools - Specialized utilities and functions

Traditional Problem:

Company has AI agent → Needs to build custom RAG system
                     → Needs to find datasets
                     → Needs to integrate APIs
                     → Needs documentation
                     → Spends weeks building infrastructure

Axiomeer Solution:

Company has AI agent → Searches Axiomeer marketplace
                     → Finds ready-to-use RAG system
                     → Discovers relevant datasets
                     → Gets API integrations
                     → Accesses processed documents
                     → Everything ready in minutes

Need: AI agent that answers customer questions about products Axiomeer Provides:

• Product documentation RAG system
• Customer interaction datasets
• FAQ knowledge bases
• CRM API integrations
• Pre-built response templates

Need: AI agent that generates insights from company data Axiomeer Provides:

• Analytics datasets
• Database MCP connectors
• Visualization APIs
• Statistical tools
• Industry benchmark data

Need: AI agent that creates marketing content Axiomeer Provides:

• Writing style datasets
• Content generation RAGs
• Image/media APIs
• SEO tools
• Template libraries

PRODUCTION MILESTONE ACHIEVED:

• 100% Product Availability (91/91 products tested and working)
• Enterprise Security (zero vulnerabilities, cryptographic secrets)
• Production Authentication (JWT + API keys + bcrypt + rate limiting)
• Docker Deployment (PostgreSQL + FastAPI + FAISS semantic search)
• Automated Testing (comprehensive health checks for all products)

Recent Achievements:

• Fixed critical rate limiter timezone bug
• Achieved 100% product availability across all 14 categories
• Enhanced semantic search for product discovery
• Added comprehensive validation and testing
• Eliminated all hardcoded secrets and security vulnerabilities

• Docker & Docker Compose
• Python 3.10+
• 4GB RAM minimum
• PostgreSQL 15+ (included in Docker)

# Clone repository
git clone https://github.com/ujjwalredd/Axiomeer.git
cd axiomeer

# Start the marketplace (includes PostgreSQL + API + Semantic Search)
docker-compose up -d

# Verify deployment
curl http://localhost:8000/health
# Expected: {"status":"ok"}

# Check available products
curl http://localhost:8000/apps | jq 'length'
# Expected: 91

# 1. Sign up
curl -X POST http://localhost:8000/auth/signup \
  -H "Content-Type: application/json" \
  -d '{
    "email": "[email protected]",
    "username": "my_ai_agent",
    "password": "secure_password_123"
  }'

# 2. Login and get access token
curl -X POST http://localhost:8000/auth/login \
  -H "Content-Type: application/json" \
  -d '{
    "email": "[email protected]",
    "password": "secure_password_123"
  }'
# Save the "access_token" from response

# 3. Create API key for your agent
TOKEN="<your_access_token>"
curl -X POST http://localhost:8000/auth/api-keys \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"name": "Production Agent Key"}'
# Save the "key" from response (starts with axm_)

# 4. Discover products for your agent's task
API_KEY="axm_xxxxx"
curl -X POST http://localhost:8000/shop \
  -H "X-API-Key: $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "task": "I need to build a RAG system for customer documentation",
    "auto_extract_capabilities": true,
    "max_results": 5
  }'

# 5. Execute a discovered product
curl -X POST http://localhost:8000/execute \
  -H "X-API-Key: $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "app_id": "wikipedia_search",
    "task": "Find information about artificial intelligence",
    "inputs": {},
    "require_citations": true
  }'

┌─────────────────────────────────────────────────────────────┐
│                     AI AGENT                                │
│  (Your custom agent needing RAGs, APIs, datasets, etc.)     │
└──────────────────────┬──────────────────────────────────────┘
                       │
                       │ 1. Task Description
                       │    "I need customer data analysis tools"
                       ▼
┌─────────────────────────────────────────────────────────────┐
│                  AXIOMEER MARKETPLACE                       │
│                                                             │
│  ┌──────────────────────────────────────────────────────┐   │
│  │  DISCOVERY ENGINE (Semantic Search + LLM)            │   │
│  │  • FAISS vector search (384-dim embeddings)          │   │
│  │  • LLM capability extraction (Ollama phi3.5)         │   │
│  │  • Weighted ranking algorithm                        │   │
│  └──────────────────────────────────────────────────────┘   │
│                                                             │
│  ┌──────────────────────────────────────────────────────┐   │
│  │  PRODUCT CATALOG                                     │   │
│  │  • RAG Systems         • Datasets                    │   │
│  │  • MCP Servers         • APIs (91+)                  │   │
│  │  • Documents           • Agent Components            │   │
│  └──────────────────────────────────────────────────────┘   │
│                                                             │
│  ┌──────────────────────────────────────────────────────┐   │
│  │  EVALUATION & RANKING                                │   │
│  │  • Capability Match (70%)                            │   │
│  │  • Semantic Relevance (25%)                          │   │
│  │  • Trust Score (15%)                                 │   │
│  │  • Latency Penalty (20%)                             │   │
│  │  • Cost Factor (10%)                                 │   │
│  └──────────────────────────────────────────────────────┘   │
└──────────────────────┬──────────────────────────────────────┘
                       │
                       │ 2. Ranked Product Recommendations
                       │    [Analytics Dataset, CRM API, ...]
                       ▼
┌─────────────────────────────────────────────────────────────┐
│                     AI AGENT                                │
│  Receives ready-to-use products matching requirements       │
└─────────────────────────────────────────────────────────────┘

1. Semantic Search (FAISS)

• 384-dimensional embeddings (all-MiniLM-L6-v2)
• Sub-100ms vector similarity search
• Finds products by meaning, not just keywords

2. LLM Capability Extraction

• Ollama phi3.5:3.8b model
• Automatically extracts required capabilities from natural language
• Example: "analyze customer feedback" → ["analytics", "nlp", "sentiment"]

3. Weighted Ranking Algorithm

score = (
    0.70 * capability_coverage +    # Must have required features
    0.25 * semantic_relevance +     # How well it matches the task
    0.15 * trust_score -            # Reliability and quality
    0.20 * latency_penalty -        # Response time impact
    0.10 * cost_factor              # Usage cost
)

• Ollama Mistral 7B - General purpose LLM
• Ollama Llama3 8B - Advanced reasoning
• Ollama CodeLlama 13B - Code generation
• Ollama DeepSeek Coder - Specialized coding

• Exchange Rates API - Real-time currency data
• Blockchain Info - Bitcoin/crypto analytics
• Coinbase Prices - Cryptocurrency pricing

• Pokemon Data - Complete Pokemon database
• Cat Facts - Animal knowledge base
• Dog Images - Image dataset API
• Breaking Bad Quotes - TV show dataset
• Rick & Morty Characters - Character database

• TheMealDB - Recipe database
• Fruityvice - Nutrition information
• CocktailDB - Drink recipes

• Chuck Norris Jokes - Humor dataset
• Kanye Quotes - Celebrity quotes
• Dice Roll - Random number generation
• Coin Flip - Decision tools
• Corporate BS Generator - Business text
• Yes/No API - Decision helper
• Useless Facts - Trivia data
• Numbers Trivia - Mathematical facts

• IP Geolocation - Location services
• REST Countries - Country information
• Nominatim Geocoding - Address lookup
• Geonames - Geographic database

• NASA APOD - Space imagery
• NASA Asteroids - Astronomy data
• NASA Mars Rover - Mars mission data
• World Bank Indicators - Economic data
• COVID Stats - Pandemic tracking
• Census Demographics - Population data
• UK Police Data - Crime statistics
• Data.gov - US government data
• EU Open Data - European datasets
• FRED Economic - Federal Reserve data
• IMF Financial - International finance

• Wikipedia Search - Encyclopedia
• Wikipedia Dumps - Full text access
• Wikidata Search - Structured knowledge
• Wikidata SPARQL - Query endpoint
• DBpedia SPARQL - Semantic web data
• PubMed Search - Medical research
• arXiv Papers - Scientific preprints
• Crossref Metadata - Academic citations
• Semantic Scholar - AI research papers
• Open Library - Book information
• Archive.org - Digital library
• Gutenberg Books - Public domain texts
• Universities Search - Educational institutions

• Dictionary API - Word definitions
• Word Synonyms - Thesaurus
• Language Detection - Auto-detect languages
• Lorem Ipsum - Placeholder text
• Gender Prediction - Name analysis
• Datamuse - Word associations

• Unsplash Photos - Stock photography
• TMDB Movies - Film database
• OMDB Movies - Movie information
• Genius Lyrics - Song lyrics
• MusicBrainz - Music metadata
• Pexels Media - Stock media

• Advice Slip - Random advice
• ZenQuotes - Inspirational quotes
• Quotable - Famous quotations

• Newton Math - Mathematical operations
• Periodic Table - Chemical elements
• Sunrise/Sunset - Solar calculations
• Space People - Astronaut tracker
• Random User Data - Person generator

• UUID Generator - Unique identifiers
• QR Code Generator - QR creation
• Base64 Encode/Decode - Data encoding
• Color Info - Color information
• Placeholder Images - Image placeholders
• HTTPBin - HTTP testing
• Postman Echo - API testing
• IP Address Lookup - IP information
• Random Data Generator - Test data
• Trivia Questions - Quiz content
• Joke API - Humor content
• Bored Activities - Activity suggestions
• Zippopotam - Postal code lookup

# Your agent describes what it needs in natural language
response = marketplace.shop(
    task="I need tools to analyze customer sentiment from reviews",
    auto_extract_capabilities=True  # LLM extracts: ["analytics", "nlp", "sentiment"]
)
# Returns: Sentiment analysis APIs, NLP datasets, review RAGs

# Execute discovered products immediately
result = marketplace.execute(
    app_id="sentiment_analyzer",
    task="Analyze this review: 'Great product!'",
    inputs={"text": "Great product!"}
)
# Returns: Structured sentiment analysis with citations

• Automatic cost tracking per product
• Latency monitoring
• Success rate analytics
• Citation and provenance tracking

• JWT token authentication (HMAC-SHA256)
• API key management (SHA-256 hashed)
• Rate limiting (tier-based: 100/1000/10000 req/hr)
• bcrypt password hashing (cost 12)
• Audit trails for all executions

Add your RAG, dataset, MCP server, or API to the marketplace:

{
  "id": "my_customer_rag",
  "name": "Customer Support RAG System",
  "description": "Pre-built RAG for customer service documentation",
  "category": "rag_systems",
  "capabilities": ["retrieval", "qa", "customer_support"],
  "product_type": "rag",
  "cost_est_usd": 0.01,
  "latency_est_ms": 200,
  "executor_type": "http_api",
  "executor_url": "https://your-server.com/rag/query",
  "test_inputs": {"query": "How do I reset my password?"}
}

• List free products for exposure
• Charge per-use for premium products
• Track usage analytics
• Build reputation through trust scores

• Framework: FastAPI (async, high-performance Python)
• Database: PostgreSQL 15 (with Alembic migrations)
• Search: FAISS vector similarity (384-dim embeddings)
• LLM: Ollama phi3.5:3.8b (capability extraction)
• Auth: JWT (python-jose) + bcrypt password hashing
• Validation: Pydantic v2 (type-safe request/response)

• Containers: Docker + Docker Compose
• Orchestration: Multi-container setup (API + PostgreSQL)
• Monitoring: Health check endpoints + automated testing
• Scaling: Stateless API design (horizontal scaling ready)

• Embeddings: sentence-transformers (all-MiniLM-L6-v2)
• Vector Search: FAISS CPU (Facebook AI Similarity Search)
• LLM Integration: Ollama API (local model execution)
• Semantic Matching: Cosine similarity scoring

AI agents describe their needs, marketplace returns ranked recommendations.

Request:

{
  "task": "I need to build a customer support chatbot with FAQ capabilities",
  "auto_extract_capabilities": true,
  "max_results": 10,
  "required_capabilities": ["qa", "retrieval"],  // Optional
  "client_id": "my_ai_agent_v1"  // Optional tracking
}

Response:

{
  "matches": [
    {
      "app_id": "faq_rag_system",
      "name": "FAQ RAG System",
      "description": "Pre-built retrieval system for frequently asked questions",
      "score": 0.95,
      "capability_coverage": 1.0,
      "semantic_relevance": 0.89,
      "why_matched": "Perfect capability match for QA and retrieval"
    }
  ],
  "total_searched": 91,
  "llm_extracted_capabilities": ["qa", "retrieval", "chatbot", "support"]
}

Execute a discovered product with specific inputs.

Request:

{
  "app_id": "faq_rag_system",
  "task": "How do I reset my password?",
  "inputs": {
    "query": "password reset procedure"
  },
  "require_citations": true
}

Response:

{
  "app_id": "faq_rag_system",
  "ok": true,
  "output": {
    "answer": "To reset your password: 1. Go to login page, 2. Click 'Forgot Password'...",
    "citations": ["https://docs.example.com/password-reset"],
    "retrieved_at": "2026-02-08T12:34:56Z",
    "confidence": 0.92
  },
  "provenance": {
    "sources": ["Internal FAQ database"],
    "method": "vector_similarity_search"
  },
  "run_id": 12345
}

# Tests all 91 products automatically
python scripts/api_health_check.py

Expected Output:

================================================================================
AXIOMEER PRODUCTION READINESS - API HEALTH CHECK
================================================================================

Loading API manifests...
Found 91 APIs to test

Testing APIs...

[1/91] Testing exchange_rates... ✓ PASS (324ms)
[2/91] Testing blockchain_info... ✓ PASS (456ms)
...
[91/91] Testing ollama_deepseek_coder... ✓ PASS (19ms)

================================================================================
SUMMARY
================================================================================

Total APIs: 91
Passed: 91 (100.0%)
Failed: 0 (0.0%)

================================================================================
PRODUCTION READINESS ASSESSMENT
================================================================================

✓ ALL APIS PASSING - PRODUCTION READY

# Run test suite
pytest tests/ -v

# Run specific category
pytest tests/test_auth.py -v
pytest tests/test_rate_limit.py -v

# Database Configuration
DB_PASSWORD=<43-char-cryptographic-secret>
DATABASE_URL=postgresql://axiomeer:${DB_PASSWORD}@localhost:5432/axiomeer

# Authentication (REQUIRED in production)
AUTH_ENABLED=true
JWT_SECRET_KEY=<43-char-cryptographic-secret>
JWT_ACCESS_TOKEN_EXPIRE_MINUTES=60

# Rate Limiting
RATE_LIMIT_ENABLED=true
RATE_LIMIT_FREE_TIER_PER_HOUR=100
RATE_LIMIT_STARTER_TIER_PER_HOUR=1000
RATE_LIMIT_PRO_TIER_PER_HOUR=10000

# Semantic Search
SEMANTIC_SEARCH_ENABLED=true
SEMANTIC_SEARCH_MODEL=all-MiniLM-L6-v2

# LLM (Ollama)
OLLAMA_URL=http://localhost:11434/api/generate
OLLAMA_MODEL=phi3.5:3.8b

• JWT tokens with HMAC-SHA256 signing
• API keys with SHA-256 hashing
• bcrypt password hashing (cost factor 12)
• Token expiration (configurable, default 60 min)

• Tier-based rate limiting
• Usage tracking per user
• API key management (create/revoke)
• Fine-grained access control

• Secure PostgreSQL configuration
• Docker network isolation
• No hardcoded secrets (environment-based)
• Fail-fast security validation

• Audit trails (all executions logged)
• Citation tracking (provenance for AI outputs)
• Cost tracking (transparent pricing)
• GDPR-ready (user data management)

• API Documentation: See API Reference
• Product Manifests: manifests/categories/
• Source Code: src/marketplace/
• Tests: tests/

Issue: "Not authenticated"

# Solution: Include API key or JWT token
curl -H "X-API-Key: axm_xxxxx" http://localhost:8000/execute ...
# OR
curl -H "Authorization: Bearer <token>" http://localhost:8000/execute ...

Issue: "Rate limit exceeded"

# Solution: Wait for rate limit window to reset (shown in error)
# OR upgrade to higher tier
# Check current limits:
curl -H "X-API-Key: axm_xxxxx" http://localhost:8000/auth/me

Issue: Products not loading

# Solution: Restart API container to reload manifests
docker-compose restart api

# Verify products loaded
curl http://localhost:8000/apps | jq 'length'  # Should return 91

• Product discovery and ranking
• 91 products across 14 categories
• Authentication and rate limiting
• Semantic search with FAISS
• Docker deployment

Want to list your RAG, dataset, MCP server, or API?

1. Create a product manifest (JSON)
2. Implement the executor interface
3. Submit via pull request
4. Pass quality verification
5. Go live on the marketplace!

Template:

{
  "id": "your_product_id",
  "name": "Your Product Name",
  "description": "Clear description of what your product does",
  "category": "rag_systems|datasets|apis|mcp_servers|documents",
  "capabilities": ["capability1", "capability2"],
  "product_type": "rag|dataset|api|mcp|document",
  "freshness": "static|daily|realtime",
  "citations_supported": true,
  "latency_est_ms": 200,
  "cost_est_usd": 0.01,
  "executor_type": "http_api|grpc|local",
  "executor_url": "https://your-endpoint.com/api",
  "test_inputs": {"param": "example_value"}
}

• Report bugs via GitHub Issues
• Submit feature requests
• Contribute code improvements
• Improve documentation

Proprietary - © 2026 Axiomeer

Contact: [[email protected]]

Built with:

• FastAPI - Modern Python web framework
• PostgreSQL - Reliable database
• FAISS - Efficient vector similarity search
• Sentence Transformers - Semantic embeddings
• Ollama - Local LLM execution
• Docker - Containerization

Made for the AI Agent community

Empowering AI agents to access the tools they need, when they need them.