Allow for multiple LLM providers - #25

Merged

ccmdi merged 25 commits into

claude/codebase-review-01KknsD8mSzzuNTdeo2RhVaa

Nov 19, 2025

Commits on Nov 18, 2025

feat: add multi-provider LLM support with litellm
claude
committed
2332024
refactor: focus on model choice instead of provider choice
claude
committed
6fdc96b
fix: use only real models that actually exist
claude
committed
449cf41
fix: add GPT-5 and Gemini 3 Pro Preview (the actual latest models)
claude
committed
236dcdb
fix: imports
ccmdi
committed
12b8601
feat: allow setting model via human-friendly names
claude
committed
3865296
docs: add example of setting model in README
claude
committed
60463b7
Merge branch 'claude/codebase-review-01KknsD8mSzzuNTdeo2RhVaa' of https://github.com/ccmdi/obsidianki into claude/codebase-review-01KknsD8mSzzuNTdeo2RhVaa
ccmdi
committed
abdd9c9
fix: lazy load cmds
ccmdi
committed
23402ea
refactor: simplify to single MODEL config (remove ai_provider/ai_model)
claude
committed
5438b6b
Merge remote-tracking branch 'origin/claude/codebase-review-01KknsD8mSzzuNTdeo2RhVaa' into claude/codebase-review-01KknsD8mSzzuNTdeo2RhVaa
ccmdi
committed
cda3cec
refactor: avoid ai model redundancy
ccmdi
committed
3da06c7
feat: centralize litellm logic
ccmdi
committed
68b9c4a
feat: setup dropdown for model selector
ccmdi
committed
b6db331
chore: model name fix
ccmdi
committed
897d663

Commits on Nov 19, 2025

fix: add questionary mocking to setup tests and fix model_choice bug
claude
committed
47db856
fix: reload command modules in mock_services to fix test mocking
claude
committed
c5bf4a6
fix: move service import to function definition for commands
ccmdi
committed
fc78d2c
test: add multi-provider feature test coverage
claude
committed
42b5d89
fix: standardize tool_choice format for different LLM providers
claude
committed
36e94e6
fix: tool choice unification
ccmdi
committed
27b6c69
fix: gemini flash 2.5 model signature
ccmdi
committed
1acaf44
fix: add proper type annotations to ai/client.py
claude
committed
a1282e1
.
ccmdi
committed
57248f6
fix: use cast() to handle ModelResponse vs CustomStreamWrapper typing
claude
committed
d9a6840