fix: support Pydantic Field in Annotated tool parameters

[MohammadHijjawi97]

Related Issues

• fixes: no existing issue (searched open/closed issues and PRs for Annotated + Field in tools, no match)

Proposed Changes:

create_tool_from_function (and therefore @tool) took the first typing.Annotated metadata item and wrote it into the JSON schema as the parameter description, whatever its type. With the common Pydantic pattern Annotated[int, Field(ge=1)] (or Annotated[str, Field(description="...")]), a FieldInfo object ended up in the schema and creating the tool failed:

from typing import Annotated
from pydantic import Field
from haystack.tools import tool

@tool
def forecast(city: str, days: Annotated[int, Field(ge=1)] = 1) -> str:
    """Get the forecast."""
    return ""
# ValueError: The provided parameters do not define a valid JSON schema

Now only a string metadata item is used as the description; other metadata is left to Pydantic, so Field constraints (minimum, minLength, ...) and a Field(description=...) show up in the tool schema. The plain Annotated[str, "description"] behavior is unchanged, and the docstring is updated accordingly.

How did you test it?

• Added test_from_function_annotated_with_pydantic_field (a Field plus a string description, a Field with its own description, and a Field without description). It fails on main with the error above and passes with the fix.
• pytest test/tools -m "not integration": 272 passed.
• ruff check / ruff format --check and mypy on the changed files.

Notes for the reviewer

When both a string and a Field(description=...) are present, the string wins, consistent with the documented Annotated[T, "description"] convention.

I used an AI coding assistant (Claude Code) while writing the fix and tests; I reviewed the changes and ran the checks listed above.

Checklist

• I have read the contributors guidelines and the code of conduct.
• I have updated the related issue with new insights and changes. (no related issue)
• I have added unit tests and updated the docstrings.
• I've used one of the conventional commit types for my PR title.
• I have documented my code.
• I have added a release note file, following the contributors guidelines.
• I have run pre-commit hooks and fixed any issue. (ran ruff and mypy on the changed files)

[vercel]

@MohammadHijjawi97 is attempting to deploy a commit to the deepset Team on Vercel.

A member of the Team first needs to authorize it.

[github-actions]

Hi @MohammadHijjawi97, thanks for your interest in contributing to Haystack! 🙏

⚠️ You currently have 3 open pull requests in this repository (#12793, #12790 and this one). Our review capacity is limited, so please hold off opening more PRs until we've had a chance to review your first 2 open PRs. This helps us give each contribution the attention it deserves. Thank you!

This is an automated message to help us keep the review queue healthy.

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
haystack/tools
from_function.py
Project Total

This report was generated by python-coverage-comment-action

[sjrl]

@MohammadHijjawi97 thanks for looking into this! Could I ask you if this is something you ran into while building tools with Haystack?

[MohammadHijjawi97]

Not in a production project, no. I came across it while going through how create_tool_from_function handles Annotated metadata and trying the Annotated[int, Field(ge=1)] pattern from the Pydantic docs, which fails at tool creation today.

[MohammadHijjawi97]

@sjrl following up on your question with a concrete case where this comes up: functions written for MCP servers. FastMCP documents parameters as Annotated[int, Field(description=..., ge=1)], so reusing such a function as a Haystack tool (create_tool_from_function / @tool) currently fails at tool creation with The provided parameters do not define a valid JSON schema. With this change the Field description and constraints (minimum, maxLength, ...) end up in the tool schema the LLM sees, and plain Annotated[T, "description"] keeps working as before. I've merged current main into the branch. @bogdankostic, it's ready for review whenever you have time.