gitCommitGenie.autoStageAllForDiff |
boolean |
false |
Only when the staging area is empty: temporarily stage all changes to build the diff, then restore your staging state. Use with caution—this may include unrelated changes in the prompt. |
gitCommitGenie.chain.enabled |
boolean |
false |
Enable multi-step Thinking mode for commit generation (more detailed and accurate, better template adherence, but higher latency and token usage). |
gitCommitGenie.chain.maxParallel |
number |
2 |
Maximum parallel LLM calls used by Thinking mode across all providers. Increase carefully to avoid provider rate limits. |
gitCommitGenie.llm.maxRetries |
number |
2 |
Max retry attempts for API request failures. |
gitCommitGenie.llm.temperature |
number |
1 |
Temperature (0–2). Default 1. Some provider/model combinations only accept 1; changing this value may trigger invalid-temperature errors or less stable outputs. |
gitCommitGenie.rag.enabled |
boolean |
false |
Enable RAG to keep commit message generation style consistent (requires sufficient historical commits to build a local style index). Configure the embedding API key first via the "Configure RAG Embedding API Key" command. |
gitCommitGenie.rag.embedding.baseUrl |
string |
"" |
Base URL for the embeddings API used by RAG. Must be compatible with OpenAI SDK format. |
gitCommitGenie.rag.embedding.model |
string |
"" |
Embedding model used by RAG. |
gitCommitGenie.rag.embedding.dimensions |
number |
0 |
Optional embedding dimensions used by RAG (0 = auto-detect). |
gitCommitGenie.rag.embedding.batchSize |
number |
10 |
Batch size used when RAG generates embeddings (1–512). |
gitCommitGenie.repositoryAnalysis.enabled |
boolean |
true |
Enable repository analysis to provide better context for commit message generation. |
gitCommitGenie.repositoryAnalysis.excludePatterns |
array |
[] |
File patterns to exclude from repository analysis scanning (gitignore-style). |
gitCommitGenie.repositoryAnalysis.updateThreshold |
number |
10 |
Number of commits after which to update the repository analysis. |
gitCommitGenie.repositoryAnalysis.MaxCount |
number |
unlimited |
Maximum number of analysis steps allowed during repository exploration. Set to -1 for unlimited steps (default). |
gitCommitGenie.repositoryAnalysis.model |
enum |
general |
Model used for repository analysis. Pick any supported model across providers; the provider automatically switches to match your selection. Choose "Use default model" to reuse your main commit message model. You can configure this via "Manage Models" command. |
gitCommitGenie.commitLanguage |
string |
auto |
Target language for generated commit messages. Options: auto, en, zh-CN, zh-TW, ja, ko, de, fr, es, pt, ru, it. |
gitCommitGenie.typingAnimationSpeed |
number |
15 |
Speed of the commit message box typing animation in milliseconds per character. Set to -1 to disable the animation. |
gitCommitGenie.showUsageCost |
boolean |
true |
When enabled, a brief notification displays the estimated total cost for the current generation. |
gitCommitGenie.ui.stageNotifications.enabled |
boolean |
true |
Show bottom‑right stage notifications during Thinking. Title‑less minimal bubble. |