Enhance autocomplete, debugging, and testing features with version updates - #25
Merged
Conversation
- Implemented OpenAiCompatibleFimClient for handling autocomplete requests. - Created settings management for autocomplete features including enabling/disabling, base URL, model, and other parameters. - Added UI components for configuring autocomplete settings in the settings panel. - Integrated autocomplete settings into the sidebar and protocol for persistence and retrieval. - Developed tests for autocomplete functionality and settings normalization.
- Introduced a new skill `debug-systematic` for systematic debugging processes. - Enhanced `userSafetyRules` in contracts to tighten user safety without widening grants. - Implemented intersecting user safety rules in decision policy to ensure stricter control over tool grants. - Added approval presets for VS Code to manage user interactions effectively. - Created utility functions for formatting effective grants and approval presets. - Added tests for new functionalities including user safety rules and effective grant formatting.
- Added new capstone tasks to the frontend benchmark suite, increasing total cases from 85 to 86. - Updated expected counts in the frontend suite to reflect the new total. - Adjusted the description in the frontend suite to accurately represent the number of cases. - Enhanced testing cases in the testing suite, increasing total cases from 23 to 24, with specific updates to easy, medium, and hard categories. - Added new checks to various test cases to ensure comprehensive coverage and validation. - Introduced new verification tests for file contents and HTTP multi-step requests to improve testing robustness.
…prove decision policy handling
Resolve version conflicts by keeping local 2.9.17 across product packages. Co-authored-by: Cursor <cursoragent@cursor.com>
…erver functionality - Introduced an HTML interface for the Mitii Log Viewer, allowing users to inspect logs, model I/O, tool calls, and errors. - Implemented a Node.js server to serve the log viewer and handle API requests for loading logs and retrieving default repository paths. - Added functionality to read and parse log files, including model I/O and session logs, with error handling and issue reporting. - Included support for file selection and directory loading through the UI.
- Added "already_read" to TOOL_REASON_CODES for better state management. - Improved mutation budget selection logic to keep scaffold-like package feature work on a "standard" profile instead of "tight". - Introduced `remapScaffoldChangeSurfaces` and `resolveScaffoldPackageMapping` functions to handle package scaffolding more effectively. - Updated `compileDiscoveryBrief` to remap scaffold template surfaces onto target packages. - Enhanced skill budget application to allow omitting large skills based on the `forbidLargeSkills` flag. - Implemented tag evidence requirements for skills to prevent unintended matches. - Added tests for new functionalities and ensured existing tests cover the changes effectively.
…ol grant logic - Added a test to ensure fast failure when no-tool routes leak workspace tool calls. - Updated existing tests to clarify failure conditions and improve messaging. - Enhanced mutation budget resolution to maintain standard profiles for scaffold and migrate tasks. - Introduced new logic for handling continuation artifact requests in decision policy. - Improved remapping of scaffold change surfaces to better handle package ports and migrations. - Updated planning actions to reject truncated discovery drafts and utilize deterministic skeletons. - Adjusted task item path matching to ensure only explicit write targets are considered. - Increased output token limits for tool-loop executions to accommodate larger budgets.
…and improve whitespace handling
- Introduced function to manage user-approved Continue overrides during budget exhaustion and unfulfilled executions. - Updated , , and to utilize the new continue mechanism. - Enhanced to track the count of Continue overrides. - Added tests for the new continue logic and updated existing tests to reflect changes in behavior from failure to suspension. - Adjusted pipeline verification and repair logic to support the new continue mechanism, allowing for user intervention on stalls.
added 3 commits
September 9, 2026 00:59
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This pull request introduces support for the new
@mitii/automationand@mitii/daemonpackages throughout the build, release, and CI/CD workflows, and adds a new unattended CI entrypoint to the CLI viamitii run --auto. It also documents and exposes the new FIM autocomplete feature, improves help and documentation for automation and safety, and adds image attachment support for CLI prompts. Several tests and supporting utilities are included for the new features.Build and Release Workflow Updates:
@mitii/automationand@mitii/daemonto the publish chain and build steps in.github/workflows/npm-publish.yml,.vscode/tasks.json, and related scripts, ensuring these packages are versioned, built, and published like the others. [1]R1, [2] [3] [4] [5]automationanddaemonpackage files.CLI: Unattended CI Automation (
run --auto):mitii run --auto "<task>"command for unattended, approval-free CI runs, with supporting argument parsing, help text, and documentation. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11]FIM Autocomplete Feature:
README.md, including configuration instructions and feature summary. [1] [2] [3]Image Attachment Support:
.png,.jpg,.jpeg,.webp, and.gifformats. [1] [2]General Improvements and Maintenance:
Let me know if you'd like a walkthrough of the new
run --autoflow or the FIM autocomplete setup!