Loading repository data…
Loading repository data…
dikkadev / repository
turbocommit is a Rust-based CLI tool that generates high-quality git commit messages in accordance with the Conventional Commits specification, using OpenAI API compatible service. It is easy to use and a cost-effective way to keep git commit history at a higher quality, helping developers stay on track with their work.
A CLI tool that uses OpenAI gpt-5.4 to generate high-quality conventional commit messages from staged changes in Git and Jujutsu (JJ) repositories.
Version 3.0 standardizes on GPT-5.4. Legacy GPT-5.1 and multi-model compatibility paths have been removed.
gpt-5.4 only, with no legacy model fallbacksnone, low, medium, highlow, medium, high~/.turbocommit.yamlcargo install turbocommit
Optional shell alias:
alias tc='turbocommit'
git add .
turbocommit
After generation you can select a suggestion, edit it, ask for revisions, or commit it directly.
turboCommit now supports only gpt-5.4.
-n <number>: number of commit message suggestions to generate, default 3-m, --model <model>: model to use, must be gpt-5.4-e, --reasoning-effort <level>: none, low, medium, high-v, --verbosity <level>: low, medium, high-d, --debug: print request and usage details--debug-file <path>: write detailed debug logs to a file, or - for stdout--auto-commit: commit automatically using the generated message--amend: regenerate the last commit message from the last commit diff--api-key <key>: provide API key directly--api-endpoint <url>: override the API endpoint-c, --config <path>: load a non-default config file-r, --revision <rev>: select a JJ revision to describe--rw: toggle JJ rewrite modegpt-5.4 supports configurable reasoning effort.
turbocommit -m gpt-5.4
turbocommit --reasoning-effort high -m gpt-5.4
turbocommit --reasoning-effort none -m gpt-5.4
turbocommit --verbosity low -m gpt-5.4
turbocommit --verbosity medium -m gpt-5.4
turbocommit --verbosity high -m gpt-5.4
turbocommit -d
turbocommit --debug-file debug.log
turbocommit --debug-file -
Debug logs include request parameters, API responses or errors, token counts, and elapsed time.
The tool is now documented against OpenAI's current GPT-5.4 API pricing.
gpt-5.4 input: $2.50 / 1M tokensgpt-5.4 cached input: $0.25 / 1M tokensgpt-5.4 output: $15.00 / 1M tokensNotes:
gpt-5.4-pro exists, but this CLI does not target it.gpt-5.4, with higher pricing for prompts above 272K input tokens.v1/chat/completions, which OpenAI documents as supported for gpt-5.4.turboCommit creates ~/.turbocommit.yaml on first run.
Example:
model: "gpt-5.4"
default_number_of_choices: 3
reasoning_effort: "low"
verbosity: "medium"
disable_auto_update_check: false
api_endpoint: "https://api.openai.com/v1/chat/completions"
api_key_env_var: "OPENAI_API_KEY"
Important:
model must be gpt-5.4system_msg values are rejected and the default prompt is shown in the validation errorturbocommit -c ./local-config.yaml
turbocommit -c ~/.turbocommit-azure.yaml
turbocommit
Use --amend when you want to improve the last commit message without staged changes.
git status
turbocommit --amend
turbocommit --amend --auto-commit
Constraints:
--amendRecommended workflow:
turbocommit --amend after checks pass if you want a better message.A disposable Dev Container is included for validating Git and JJ integration without touching real repositories.
devcontainer up --workspace-folder .
devcontainer exec --workspace-folder . bash
Issues and pull requests are welcome.
Licensed under MIT. See LICENSE.