CodeGryphon’s history since 2024, recent releases and plans. Released features and future work are listed separately.
Updated
Built the Desktop foundation: a workspace connected to the files of a specific project.
Created a persistent workspace for everyday development.
Added browser access to projects and conversations.
Added collaboration tools and shared project rules.
Expanded context beyond the prompt to files, images and accumulated decisions.
Refined the application for different operating systems and parallel projects.
Connected shell commands to the project workspace.
Structured complex work into tasks with sequential steps and visible progress.
Improved synchronization and selection of the files Gryphon works with.
Brought task management beyond the Desktop window and made command execution easier to follow.
Added external connections and reusable scenarios with project-level settings.
Expanded Git integration and project-level execution mode selection.
Added automatic mode and expanded usage and application update controls.
Separated work streams within a project and introduced personalized interaction settings.
Added recurring checks and tasks that run in project context.
Embedded a browser so Gryphon can check interfaces and present the results.
Brought project health and recurring work into a shared Desktop panel.
Expanded Pulse controls and improved task and chat interactions.
Updated browser tools and Desktop installation, including macOS.
Published the SDLC benchmark methodology and model comparison to inform model selection for Gryphon modes.
Connecting Pulse findings, investigation in a work chat and a follow-up check into one workflow.
Refining tools and background work so execution preserves the right project, context and permissions.
Gryphon extracts a useful working method from an accepted result and offers to save it as a recipe.
Start an application visually and continue the same project in the main Gryphon workspace.
We plan to train and develop our own model for engineering tasks and working with project context.
We plan an ML mechanism that analyzes project changes and flags potential problems early.
The next product directions. Timing follows prototype validation and scope decisions.
Keep docs, rules and agreed decisions for use in later tasks.
Show code changes and command results so the team can accept or reject the work.
Develop recurring checks and tasks with explicit rules for when they run.