Release notes - Self-Hosted versions
Last updated
Was this helpful?
Last updated
Was this helpful?
New
Integration with GitHub Copilot, with the generation of .
Outdated programs are now excluded from the detection.
Fixed
Rendering issue for Swift practices
Practice generation issue for C++ practices
Good bye Packmind, welcome Packmind AI .
This release brings a new era for Packmind. Read the full announcement .
Packmind AI comes with new images from the packmind namespace on the Docker Hub. You can migrate the image names if you have already deployed Packmind.
The release comes with a new versioning starting from 1.0.0.
promyze/promyze-api
packmind/packmind-api
promyze/promyze-proxy
packmind/packmind-proxy
promyze/promyze-suggestions
packmind/packmind-detection
promyze/packmind-aiagent
packmind/packmind-ai-agent
New
Manual creation of practice suggestions is now available. From your Inbox, you can now create a suggestion manually (New -> New Practice).
Outdated detection status: When you update coding practices examples, or unit tests, the current program is run against this new configuration. If any of the examples or unit tests fail, then the practice is marked as "OUTDATED." You must generate a new program to cover the new examples and unit tests.
Removed
The Driver role has been removed. A new permission system is under construction and will be available soon.
Battles have been removed. We're on our way to designing a new experience for challenging the status quo in teams. Note that:
Practices that were in Rejected state have been removed.
Practices in Ongoing state have been prefixed by [TO DISCUSS] and their detection disabled.
Practice review has been removed and replaced by the Inbox.
Fixed
Settings are not available anymore for users who are not either Driver or Admin.
New
The practice review is marked as deprecated since it will be removed soon, as it's going to be replaced by the new Inbox experience, currently in Beta with restricted access.
Since the latest versions of the IDE plugins, it's no more possible to submit new Questions/Topics. We decided to remove this feature as we focus our effort and Packmind value on capturing and enforcing teams' technical decisions.
Fixed
Fixed an issue that occurred occasionally when running the Packmind CLI.
Fixed
Fixed an issue with Git repositories selection
Fixed
Fixed an issue that happened when user tried to onboard themselves.
Improved
The Data Board has been redesigned to highlight more effectively the impact and benefits of using Packmind
The AI Agent now includes in its validation process all the compliant and non-compliant examples generated in the detection guidelines. If you update these examples, they will be included in the validation process.
New
New view for the practices, available within the practice modal when you click on Open page.
Practices are automatically configured by AI once they are validated
Fixed
Fixed glitches in the practices list
Onboarding workshops
Reviews generated by AI
New IDE experience: introducing Tech Coach
The Tech Coach provides code reviews based on your practices and new practice suggestions straight from your IDE
Code reviews are now computed only on modified files to avoid any disruptions when coding
Code reviews and suggestions come with code correction to help making your code compliant with your practices/the suggestions
-> Here is the result for the JetBrains extension:
Practice detection "quick-configuration": AI-assisted configuration to make your practices detectable in your code, without having to deal with regular expressions of Semgrep scripts
Improved
Practices management
Ownership: a practice can be assigned to any teammate belonging in the current space.
New filters of practices page: by status ("validated", "not validated", "battle",...), by configuration status ("Not configured", "AI only", "Configured",...) and by owner.
The number of comments on a practice is now displayed in the list.
Practice review
Navigation bar displays practices name and questions to ease user experience.
Fixed
Rejected practices were eligible for the "practice of the day" scheduler, they are now excluded.
Fix UI issues with Microsoft Teams notifications sent from WebHooks.
Practices with status To validate are excluded in the violations detection process. This means that, when you create a new coding practice, you won't get notifications in your IDE regarding violations, until this practice is validated through a practice review.
Similarly, practices rejected during past battles were incorrectly included in the violation detection.
An UI issue to access catalogs from the practices page has been fixed.
New
Packmind now integrates with Git repositories hosted on Bitbucket Data Center.
It's now possible to ask AI to generate a new description for a coding practice.
Improved
The quality of the results generated by AI (coding practices and questions) has been improved for OpenAI and Azure OpenAI.
Several minor improvements within the practice review module
Fixed
Fix an issue with the validation of Semgrep patterns when Packmind runs in offline context
This is the first release of Packmind with its new identity (previously known as Promyze).
A completely rebranded UX and UI
A dedicated module for AI suggestions including a smoother navigation in the results
Improve logs on Git integrations and Azure OpenAI integrations.
Fix an issue with the Docker container that runs Packmind in a single container
Packmind leverages AI capabilities to serve generative coding practices, assisting Tech leaders with suggestions on how to improve code and question the technical choices made by developers.
Packmind AI offers continous generation of coding practices and discussion topics using the integration with Git repositories. Packmind AI scans your latest code changes to suggest new material for your next practice review workshop. This AI analysis can also be run within our IDE plugins on a specific file.
Here is an example of generad practice, that drivers must validate or not for the next practice review:
Packmind AI will also generate textual description for your coding practices along with regular expressions for their automatic detection.
Behind the scenes, Packmind AI relies on one of the following LLM:
OpenAI managed by Packmind (Cloud only)
Your own OpenAI instance
Your Azure OpenAI instance
Note that Packmind AI is not enabled by default, you should first enable it. Also, it's possible to disable AI for specific spaces to meet your legal constraints.
If your coding practices must be illustrated with code examples scattered on different files, multi-file examples come to the rescue. They are particularly relevant for coding practices related to design or architecture. You can attach a series of code examples and add a textual description to each single example.
Here is an illustration of a coding practice with two positive examples:
Packmind extension for VSCode: 8.1.1 or higher
Packmind extension for JetBrains: 7.1.0 or higher
Packmind extension for VS: 5.2.0 or higher
Packmind extension for Eclipse will be released soon by the end of 2023.
The hour of the day at which the Git scheduler runs can be tuned through the GIT_AI_SCHEDULER_HOUR environment variable on promyze-api container (use a number between 0 and 23).
The hour of the day at which the practice scheduler runs can be tuned through the PRACTICE_SCHEDULER_HOUR environment variable on promyze-api container (use a number between 0 and 23).
It's now possible to test the emails sent through the practice schedulers.
The public REST API route that retrieves the coding practices for a space now returns the categories in plain text instead of the ID references.
Several vulnerabilities (CVE) have been fixed.
Drivers stand for Tech Lead, Lead Dev, or anyone organizing and leading the practice review workshops. They are expected to be core contributors in Packmind for coding practices and discussion topics. Other developers will be Contributors, meaning they're expected to submit contributions occasionally.
You can now connect your GitLab, GitHub, Azure DevOps, and BitBucket platform to get an overview of recent code modifications and to identify coding practices and discussion topics for the next practices review workshop.
The review preparation module lists recent pieces of code and offers many filters to tune the scope (files, authors, date, …). You can quickly generate material for the next practice review workshop in a few minutes.
Each space can now configure a scheduler that will push coding practices directly to users in the space, either daily, weekly, or monthly. Packmind can automatically selects a practice to send, but Drivers can also feed the queue with specific practices. Think of this feature as an internal newsletter or reminders sent to the team. Here is an example of a notification sent on Slack:
The Data Board has been redesigned to provide a more precise overview of how your teams use Packmind and what benefits you get from it.
If you have not yet deployed Packmind, you should check this new .
If you have already deployed Packmind, you can continue the previous and upgrade to Chart 3.0.0, but make sure you use the latest images.
New Inbox experience: The Inbox becomes a unique place to review each new coding practice suggested by Packmind AI (based on code review or documentation analysis) or created manually from the IDE plugins or the web interface. This inbox replaces the practice review. Read more .
Unit tests for practice detection: You can now maintain a test suite for each coding practice to improve the precision of the detection program generated by Packmind AI. Read more .
Import & export of practices: We've introduced a new way to export and import practices in Packmind. From the Practices page, you can select the practices to export and download them as a JSON file. This is useful for transferring practices from one Packmind instance to .
Meet the new Inbox experience. The inbox is the new central place to review new coding practices generated by Packmind AI from code reviews and static content analysis, as well as the ones you've manually created. Read more on this Inbox .
Code review analysis: Packmind can integrate with your Git platform to analyze your Pull/Merge requests discussions and extract coding practices. Gitlab and GitHub are supported, BitBucket and Azure DevOps will be removed soon. Find out how to get started .
Static content import: You can now import any textual documentation (wiki, markdown), and let Packmind extract best practices from this. See more .
Inclusion/Exclusion scopes for practices detection. It is now possible to refine the scope of files/directories where violations of practices should be detected. These settings are available at the practice-level. Read more .
This release brings changes on how to deploy Packmind with Docker-Compose and Kubernetes. -> If you use Docker-Compose, check this . -> If you use Kubernetes, check our latest in 1.2.0.
Make sure you've reviewed the changes before upgrading.
Github Copilot integration: generate code complying to your practices. Using the Packmind IDE plugin, you can import your Packmind practices in a local file and allows GitHub Copilot to interact with it. Check the dedicated .
You can now authenticate to Azure Open AI using Microsoft Entra MSAL (Microsoft Authentication Library), which brings a higher level of security. Authentication using Api Key is still possible. Read the documentation .
Single Sign-On: It is now possible to enable SSO with a range of supported systems such SAML V2, Google Workspace, Microsoft Entra (formerly Azure Active Directory), Ping One, Okta, and more. Connections and data all remain local to your infrastructure. Read the documentation . This feature is available only for the Enterprise version. Reach us to get more information on how to deploy it.
It's now possible to copy practices from a space to another, which include all the source code examples. See more .
In addition to existing OpenAI and Azure OpenAI integrations, Packmind now supports OpenAI API-compatible inference systems such as or .
You can get started with Packmind AI by reading .
Multi-file examples can be managed directly through the UI using or use the following Packmind IDE extensions (read the associated doc ):
Each user is now either a in Packmind. You'll be asked to choose your role on your next sign-in.
Learn how to to Packmind in 1 min.
Learn how to enable the scheduler .
If you already have documented existing coding practices in a wiki tool (such as Confluence or Notion), you can now use the latest Packmind Web browser plugin to import them into Packmind. Read more on the .
We now offer a module to export your coding practices into a single Markdown file. Read more .
