Installation¶
Requirements¶
- Python 3.10+
- Optional: Redis for distributed cache (fallback file cache included)
Install from PyPI¶
pip install toolweaver
Optional extras (choose what you need)¶
Use extras to pull in integrations without bloating installs.
| Extra | Includes | When to use | Why |
|---|---|---|---|
toolweaver[azure] |
Azure AI Vision, Identity | Using Azure Computer Vision tools | Azure CV SDK for image analysis |
toolweaver[openai] |
OpenAI Python SDK | Using OpenAI models | GPT-4, ChatGPT API access |
toolweaver[anthropic] |
Anthropic SDK | Using Claude models | Claude 3, 3.5 API access |
toolweaver[redis] |
Redis client deps | You have Redis for distributed cache | Faster shared caching vs file cache |
toolweaver[vector-db] |
Qdrant client deps | You use semantic search/vector DB | Better tool retrieval at scale |
toolweaver[monitoring] |
WandB, Prometheus deps | You need metrics/export | Observability for prod |
toolweaver[all] |
All above | You want everything preinstalled | One-shot setup |
Examples:
pip install toolweaver[openai] # Just OpenAI
pip install toolweaver[azure,anthropic] # Multiple providers
pip install toolweaver[all] # Everything
Env hints
- Azure: AZURE_CV_ENDPOINT=..., AZURE_CV_KEY=...
- OpenAI: OPENAI_API_KEY=...
- Anthropic: ANTHROPIC_API_KEY=...
- Redis: REDIS_URL=redis://... (or rediss:// for TLS)
- Vector/Qdrant: QDRANT_URL=..., QDRANT_API_KEY=...
- Metrics: ANALYTICS_BACKEND=prometheus|otlp|sqlite, plus backend-specific envs
Verify¶
python - <<"PY"
from orchestrator import get_available_tools
print("Tools loaded:", len(get_available_tools()))
PY