Skip to content

Releases: spetr/mcp-codewizard

v0.2.10

Choose a tag to compare

@github-actions github-actions released this 04 Jan 16:34

Full Changelog: v0.2.9...v0.2.10

v0.2.9

Choose a tag to compare

@github-actions github-actions released this 04 Jan 15:24

Full Changelog: v0.2.8...v0.2.9

v0.2.8

Choose a tag to compare

@github-actions github-actions released this 03 Jan 12:41

Full Changelog: v0.2.7...v0.2.8

v0.2.7

Choose a tag to compare

@github-actions github-actions released this 03 Jan 11:52

Full Changelog: v0.2.6...v0.2.7

v0.2.6

Choose a tag to compare

@github-actions github-actions released this 03 Jan 11:29

Full Changelog: v0.2.5...v0.2.6

v0.2.5

Choose a tag to compare

@github-actions github-actions released this 03 Jan 10:38

Full Changelog: v0.2.4...v0.2.5

v0.2.4

Choose a tag to compare

@github-actions github-actions released this 01 Jan 17:22

What's Changed

Bug Fixes

Fixed OpenAI API key not being loaded from config - The configuration structs only had mapstructure tags which are used for reading YAML, but Viper uses go-yaml for writing which requires yaml tags. This caused fields like api_key to be written as apikey and then not found when reading back.

If you created a config with v0.2.3 and OpenAI provider, you need to either:

  1. Run mcp init again to regenerate the config
  2. Or manually change apikey: to api_key: in your .mcp-codewizard/config.yaml

Full Changelog

v0.2.3...v0.2.4

v0.2.3

Choose a tag to compare

@github-actions github-actions released this 01 Jan 16:56

What's Changed

Improved Setup Wizard

The mcp init wizard has been completely redesigned with a step-by-step guided setup:

Step 1: AI Provider Selection

  • Auto-detects available providers (Ollama, OpenAI)
  • Shows connection status for each provider
  • Supports custom OpenAI-compatible endpoints (Azure OpenAI, etc.)

Step 2: Endpoint Configuration

  • Tests connection to selected provider
  • Validates API keys for OpenAI
  • Shows clear error messages if connection fails

Step 3: Model Selection

  • Queries available models from the connected provider
  • Shows embedding models with size information
  • Recommends optimal model (nomic-embed-code for Ollama)

Step 4: Reranker Configuration

  • Optional reranker setup for improved search accuracy
  • Shows available reranker models from Ollama
  • Can be disabled to save resources

Other Changes

  • Fixed local registration to use .mcp.json instead of .claude.json
  • Updated documentation with new wizard examples

Full Changelog

v0.2.2...v0.2.3

v0.2.2

Choose a tag to compare

@spetr spetr released this 01 Jan 01:36

Changes

  • Add static linking for Windows builds
  • Fix OpenAI provider to support custom endpoint (BaseURL)
  • Fix Windows AMD64 build to use clang instead of gcc
  • Improve PATH handling in installers
  • Fix local registration to use .mcp.json instead of .claude.json

v0.2.1

Choose a tag to compare

@github-actions github-actions released this 01 Jan 00:29

Full Changelog: v0.2.0...v0.2.1