Skip to content

Swih/mistral-mcp

v0.5.0 Breaking

This release includes breaking changes for platform teams planning a safe upgrade.

Published 1mo MCP Developer Tools
✓ No known CVEs patched
Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

agent-skills agent-tools claude claude-code claude-desktop codestral
+14 more
cursor document-ai llm llm-tools mcp mcp-server mcp-servers mistral mistral-ai model-context-protocol ocr typescript voxstral zed

Summary

AI summary

Added deterministic seed control, structured response formats, OCR annotation features, and Docker image support.

Full changelog

## [0.5.0] - 2026-04-28

Added

  • Docker image support: multi-stage Dockerfile and .dockerignore for docker run -i --rm usage with MCP stdio clients.
  • mistral_ocr can request Mistral Document AI annotations through document_annotation_format, bbox_annotation_format, and document_annotation_prompt.
  • mistral_ocr now exposes document_annotation, per-image image_annotation, optional grouped annotations, and OCR confidence scores in structuredContent.
  • seed input parameter on mistral_chat, mistral_chat_stream, mistral_tool_call, and codestral_fim for deterministic sampling — maps to the SDK's randomSeed. Source: Mistral chat completion API.
  • response_format input parameter on mistral_chat, mistral_chat_stream, and mistral_tool_call supporting text, json_object, and strict json_schema modes for structured outputs. Source: Mistral structured outputs.
  • reasoning_content field in mistral_chat / mistral_chat_stream structuredContent: Magistral models now expose their reasoning trace separately from the visible answer, parsed from the SDK's ThinkChunk / TextChunk content array. Source: Mistral reasoning models.
  • New shared helpers toSdkResponseFormat (snake_case ↔ camelCase translator) and extractTextAndReasoning (Magistral content splitter) in src/shared.ts.

Changed

  • README and README.fr realigned to the v0.5.0 development surface and Docker usage.
  • ChatSamplingParams (shared) now includes seed so every chat-style tool that spreads it picks the parameter up.

Notes

  • This is a SemVer minor bump: new optional inputs and output fields, no breaking changes.
  • Live test added for end-to-end response_format: json_schema against the Mistral API; runs only when MISTRAL_API_KEY is set.
View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track Swih/mistral-mcp

Get notified when new releases ship.

Sign up free

About Swih/mistral-mcp

MCP server exposing 22 Mistral AI capabilities (chat, OCR, audio, vision, agents, embeddings, moderation, classification, files, batch) with dual transport (stdio + Streamable HTTP), structured outputs on every tool, and 6 curated French/English prompts with argument completion. Vendor API key required.

All releases →

Related context

Beta — feedback welcome: [email protected]