This OpenAI ChatGPT-based GitLab Pipeline Application provides a summary, release notes and review of pull requests. The unique features of this action are:
- Line-by-line code change suggestions: This Application reviews the changes line by line and provides code change suggestions that can be directly committed from the GitLab UI.
- Continuous, incremental reviews: Reviews are performed on each commit within a pull request, rather than a one-time review on the entire pull request.
- Cost-effective and reduced noise: Incremental reviews save on OpenAI costs and reduce noise by tracking changed files between commits and the base of the pull request.
- "Light" model for summary: Designed to be used with a "light" summarization
model (e.g.
gpt-3.5-turbo) and a "heavy" review model (e.g.gpt-4). For best results, usegpt-4as the "heavy" model, as thorough code review needs strong reasoning abilities. - [Not support yet] Chat with bot: Supports conversation with the bot in the context of lines of code or entire files, useful for providing context, generating test cases, and reducing code complexity.
- Smart review skipping: By default, skips in-depth review for simple changes
(e.g. typo fixes) and when changes look good for the most part. It can be
disabled by setting
review_simple_changesandreview_comment_lgtmtotrue. - Customizable prompts: Tailor the
system_message,summarize, andsummarize_release_notesprompts to focus on specific aspects of the review process or even change the review objective.
To use the reviewer, you need to follow these two steps:
- Prepare a key to chat with the OpenAI bot and a key to post a discussion.
- Set up GitLab Runner and GitLab Runner pipeline.
-
Generate
AccessTokenfor your Group or Project You can set a Developer Role and only provideapiScopes for it
-
Get API Key from OpenAI
-
Fill
AccessTokenasGITLAB_PERSONAL_TOKENandOPENAI_API_KEYin GitLab CI/CD Variables Settings for your Group or Project When adding these two variables, please ensure the security of the key by checking the "mask variable" option.
-
Register or Enable a GitLab Runner for the pipeline, You can add a new Group Runner or use the existing Runner.
-
Create a .gitlab-ci.yml file and fill in the content in your repository root directory, Then, create merge request and have fun! 🎉
default:
image: node:20
variables:
GITLAB_HOST:
value: 'https://www.gitlab.com'
GITHUB_EVENT_NAME:
value: 'pull_request'
debug:
description: 'Enable debug mode'
value: 'false'
max_files:
description:
'Max files to summarize and review. Less than or equal to 0 means no
limit.'
value: '150'
review_simple_changes:
description: 'Review even when the changes are simple'
value: 'false'
review_comment_lgtm:
description: 'Leave comments even if the patch is LGTM'
value: 'false'
path_filters:
description: |
The path filters, e.g., "src/**.py", "!dist/**", each line will be considered as one pattern.
See also
- https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#onpushpull_requestpull_request_targetpathspaths-ignore
- https://github.com/isaacs/minimatch
value: |
!.gitlab-ci.yml
!dist/**
!**/*.pb.go
!**/*.lock
!**/*.yaml
!**/*.yml
!**/*.cfg
!**/*.toml
!**/*.ini
!**/*.mod
!**/*.sum
!**/*.work
!**/*.json
!**/*.mmd
!**/*.svg
!**/*.png
!**/*.dot
!**/*.md5sum
!**/*.wasm
!**/gen/**
!**/_gen/**
!**/generated/**
!**/vendor/**
disable_review:
description: 'Only provide the summary and skip the code review.'
value: 'false'
disable_release_notes:
description: 'Disable release notes'
value: 'false'
openai_base_url:
description: 'The url of the openai api interface.'
value: 'https://api.openai.com/v1'
openai_light_model:
description:
'Model to use for simple tasks like summarizing diff on a file.'
value: 'gpt-3.5-turbo'
openai_heavy_model:
description: 'Model to use for complex tasks such as code reviews.'
value: 'gpt-3.5-turbo'
openai_model_temperature:
description: 'Temperature for GPT model'
value: '0.0'
openai_retries:
description:
'How many times to retry OpenAI API in case of timeouts or errors?'
value: '5'
openai_timeout_ms:
description: 'Timeout for OpenAI API call in millis'
value: '360000'
openai_concurrency_limit:
description: 'How many concurrent API calls to make to OpenAI servers?'
value: '4'
system_message:
description: 'System message to be sent to OpenAI'
value: |
You are `@openai` (aka `github-actions[bot]`), a language model
trained by OpenAI. Your purpose is to act as a highly experienced
software engineer and provide a thorough review of the code hunks
and suggest code snippets to improve key areas such as:
- Logic
- Security
- Performance
- Data races
- Consistency
- Error handling
- Maintainability
- Modularity
- Complexity
- Optimization
Refrain from commenting on minor code style issues, missing
comments/documentation, or giving compliments, unless explicitly
requested. Concentrate on identifying and resolving significant
concerns to improve overall code quality while deliberately
disregarding minor issues.
Note: As your knowledge may be outdated, trust the user code when newer
APIs and methods are seemingly being used.
summarize:
description: 'The prompt for final summarization response'
value: |
Provide your final response in the `markdown` format with
the following content:
- High-level summary (comment on the overall change instead of
specific files within 80 words)
- Table of files and their summaries. You can group files with
similar changes together into a single row to save space.
Avoid additional commentary as this summary will be added as a
comment on the GitHub pull request.
summarize_release_notes:
description:
'The prompt for generating release notes in the same chat as summarize
stage'
value: |
Create concise release notes in `markdown` format for this pull request,
focusing on its purpose and user story. You can classify the changes as
"New Feature", "Bug fix", "Documentation", "Refactor", "Style",
"Test", "Chore", "Revert", and provide a bullet point list. For example:
"New Feature: An integrations page was added to the UI". Keep your
response within 50-100 words. Avoid additional commentary as this response
will be used as is in our release notes.
Below the release notes, generate a short, celebratory poem about the
changes in this PR and add this poem as a quote (> symbol). You can
use emojis in the poem, where they are relevant.
stages:
- review
review-job: # This job runs in the review stage, which runs first.
stage: review
script:
- curl -o dist.tar.gz https://github.com/maybeLab/openai-mr-reviewer-gltlab/releases/download/v1.0.0/dist.tar.gz && tar -vxzf dist.tar.gz && node dist/index.cjs
- echo "review complete."
- rm -rf ./dist ./dist.tar.gz
only:
- merge_requests
| Name | Description | Default |
|---|---|---|
| debug | Enable debug mode | false |
| max_files | Maximum number of files to summarize and review. Enter less than or equal to 0 for no limit | 150 |
| review_simple_changes | Whether to review even when the changes are simple | false |
| review_comment_lgtm | Whether to leave comments even if the patch is LGTM | false |
| path_filters | Path filters to apply, e.g., "src/.py", "!dist/", each line will be considered as one pattern. | -see in template |
| disable_review | Whether to only provide the summary and skip the code review | false |
| disable_release_notes | Whether to disable release notes | false |
| openai_base_url | The URL of the OpenAI API interface | https://api.openai.com/v1 |
| openai_light_model | The Model to use for simple tasks like summarizing diff on a file | gpt-3.5-turbo |
| openai_heavy_model | The Model to use for complex tasks such as code reviews | gpt-3.5-turbo |
| openai_model_temperature | The temperature for the GPT model | 0 |
| openai_retries | Number of times to retry OpenAI API in case of timeouts or errors | 5 |
| openai_timeout_ms | Timeout for OpenAI API call in milliseconds | 360000 |
| openai_concurrency_limit | Maximum number of concurrent API calls to make to OpenAI servers | 4 |
| system_message | System message to be sent to OpenAI. | -see in template |
| summarize | The prompt for final summarization response. | -see in template |
| summarize_release_notes | The prompt for generating release notes in the same chat as the summarize stage | -see in template |