GeekCode v1.0.6

GeekCode v1.0.6

What's Changed

Fixed

• /models now queries Ollama for actually installed local models instead of showing a hardcoded list
• /model <name> no longer silently routes to the wrong provider — ambiguous model names now require explicit provider/model format
• Provider inference (_infer_provider) raises clear error for ambiguous model names (llama, qwen, mixtral, etc.) instead of guessing wrong

Added

• Real-time command suggestions via prompt_toolkit — dropdown appears as you type / with descriptions for each command
• Model name autocompletion after /model — shows dynamic Ollama models and static catalog, all in provider/model format
• prompt_toolkit added as a core dependency
• CHANGELOG.md — release notes now sourced from changelog instead of hardcoded in CI

Changed

• Replaced readline tab-completion with prompt_toolkit PromptSession for the interactive REPL
• One-time automatic migration of readline history file to prompt_toolkit format
• GitHub Actions release workflow reads "What's Changed" from CHANGELOG.md

---

Installation

macOS (Homebrew):

brew install sur950/geekcode/geekcode

macOS (direct download — Apple Silicon):

curl -fsSL https://github.com/sur950/GeekCode/releases/download/v1.0.6/geekcode-macos-arm64 -o /usr/local/bin/geekcode && chmod +x /usr/local/bin/geekcode

Linux:

curl -fsSL https://github.com/sur950/GeekCode/releases/download/v1.0.6/geekcode-linux-x64 -o /usr/local/bin/geekcode && chmod +x /usr/local/bin/geekcode

Windows (Winget):

winget install sur950.GeekCode

pip:

pip install geekcode

---

Uninstall

macOS (Homebrew):

brew uninstall geekcode
brew untap sur950/geekcode

macOS / Linux (Direct Binary):

sudo rm -f /usr/local/bin/geekcode
# or if installed to custom location:
rm -f $(which geekcode)

Windows (Winget):

winget uninstall sur950.GeekCode

pip:

pip uninstall geekcode

To remove project-level data:

rm -rf .geekcode