[Idea] Adding Anthropic as a provider
Hi, great project. I think it would be beneficial to have Anthropic/Claude as a provider, in addition to Gemini and openAI. Anthropic has the best performance on code-adjacent tasks if memory serves. If this is in line with your desires, please permit me to introduce the changes required.
Thanks!
Sounds really good! I wonder if Claude api has a way to force output to a structured format like OpenAI does? I think it's quite essential for the tool to work reliably
Hmm . I think this could do: https://docs.anthropic.com/en/docs/build-with-claude/tool-use#controlling-claudes-output
Hello, Anthropic user here, I made a PR (#272) to integrate it as a plugin for humanify. I created a tool to force the output structure as you suggested @jehna.
I think this could do: docs.anthropic.com/en/docs/build-with-claude/tool-use#controlling-claudes-output
Also just saw this page:
- https://docs.anthropic.com/en/docs/build-with-claude/tool-use/token-efficient-tool-use
-
Token-efficient tool use (beta) The upgraded Claude 3.7 Sonnet model is capable of calling tools in a token-efficient manner. Requests save an average of 14% in output tokens, up to 70%, which also reduces latency. Exact token reduction and latency improvements depend on the overall response shape and size.
-
Token-efficient tool use is a beta feature. Please make sure to evaluate your responses before using it in production.
-
If you choose to experiment with this feature, we recommend using the Prompt Improver in the Console to improve your prompt.
- https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/prompt-generator
- https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/prompt-improver
-
Token-efficient tool use does not currently work with disable_parallel_tool_use.
-
To use this beta feature, simply add the beta header
token-efficient-tools-2025-02-19to a tool use request withclaude-3-7-sonnet-20250219. If you are using the SDK, ensure that you are using the beta SDK withanthropic.beta.messages.
-
See also:
- https://github.com/jehna/humanify/issues/400