- Feat: Add configurable two-column ai-code transient menu
- Chore: Add Claude Code support for Emacs MCP server
- Feat: Add an Emacs MCP server with built-in tools and agent integration for Codex CLI and GitHub Copilot CLI
- Chore: Add Investigate Issue Mode for Reviews
- Feat: Sync ECA integration with latest upstream bridge/ext changes: ai-code-menu integration, workspace management, auto project/session syncing, shared context helpers, and session health utilities
- Feat: Improve ECA backend: smarter chat-buffer handling, graceful resume, context management (file/clipboard/cursor/repo-map), session multiplexing via eca-ext.el, upgrade and install-skills support, contributed by davidwuchn
- Fix: Block ai-code-apply-prompt-on-current-file for eca backend (like agent-shell)
- Fix: Avoid eat reflow suppression on session buffers, by Silex
- Feat: Enable shared multiline terminal input in Emacs sessions and GitHub Copilot CLI compatibility, reported and verified by EricZJonsson
- Feat: Add ECA (https://eca.dev/) as an external backend, contributed by davidwuchn
- Chore: Add completion and open existing task files, also support scratch.org, for general task
- Feat: Integrate with magit worktree feature
- Feat: Modernize PR review tool
- Feat: Provide Red + Green + Blue solution in code change follow up and tdd-cycle as one command
- Feat: Add menu item to install skills for current selected AI coding CLI; with per-backend :install-skills support and generic AI-prompt fallback
- Fix: Wrong number of arguments when calling zero-arg resume functions (e.g. claude-code-ide-resume) via ai-code-cli-resume, reported and fix suggested by bbingju
- Chore: Red stage allow write test for function under cursor, add test coverage for new TDD helper functions
- Chore: Add explicit no-test prompt in auto test choices
- Fix: region-start-line missing, by limon
- Fix: ai-code-prompt-send-block
- Feat: Add repo-local backend affinity to persist AI code backend per Git repo
- Fix: Fix session lookup collisions for same-named projects, by Silex
- Fix: Preserve original whitespace in ai-code–preprocess-prompt-text, reported and fix suggested by lesteral
- Chore: Add git branch name to task file template
- Chore: Add AGENT as drawer to .ai.code.prompt.file.org
- Chore: When ai-code-use-gptel-classify-prompt is not nil, and ai-code-auto-test-type is not nil, use gptel to determine if the prompt is about to change code
- So that prompt for code reading / explaining won’t trigger test follow up; test should only be triggered when the code got changed.
- Fix: correct auto-mode-alist for ai-code-prompt-mode
- Fix: Not to kill-buffer when CLI exits on error, reported and fix suggested by lesteral
- Feat: Add agent-shell as external backend
- Fix: Always normalize git path
- Chore: Replace prefix arg with completing-read menu in context action
- Chore: Add option: ask-me, to ai-code-auto-test-type
- Feat: Create file or dir name with gptel, with ai-code-create-file-or-dir function
- Chore: Add option to create task folder, instead of single task .org file
- Chore: Allow selecting target directory for new task file creation
- Chore: Add error handling instructions to test and build commands
- Chore: Add shell command history variable and use it in prompt
- Chore: Honor prefix args for Resume AI CLI , by Silex
- Fix: Fix offscreen response notifications and ignore output noise, by Silex
- Chore: Add dedicated CLI args history, by Silex
- Fix: Stabilize backend dispatch and clean up design mismatches, by Silex
- Feat: Add agent-file property and command to open backend agent files
- Fix: Handle github copilot cli prompt sent truncate issue
- Feat: Add ai-code-auto-test-type selection to optionally let AI test the code after changes
- Chore: Let AI build / test the project, and get output as context
- Feat: Add Aider CLI support as a new AI backend integration
- Chore: Add session id field to task file template, for tracking / recovering purpose
- Chore: Enhance candidate filepath list for @ character
- visible files logic support out of repo file
- recent focused buffer file or dir in candidate list
- Feat: In AI coding session buffers, SPC in Evil normal state now triggers the prompt-entry UI
- Feat: Add CodeBuddy Code CLI backend support, by liaohanqin & CodeBuddy
- Feat: Add # class / function name completion in AI prompts, comments and AI coding session
- Chore: Include Dired directories in current frame for Git root paths list
- Feat: Enable @ completion in ai coding session window
- Feat: @ in ai-code-prompt-mode will trigger filename completion
- It will also work when ai-code-toggle-filepath-completion is triggered in config
- Feat: Add command to open recently modified files in Git repo or directory
- Chore: Better error handling when AI coding CLI failed to start
- Fix notification after visible response, by Silex
- Chore: Enhance backend selection to prioritize current choice in completion list
- Chore: Enhance ai-code-send-command to support multi-level context inputs
- Fix: Start AI CLI with Kiro-cli backend leads to error, reported by texastony
- Fix: Silence notifications-notify native compile warning, by Silex
- Feat: Add ai-code-claude-code.el, a built-in Claude Code backend using ai-code-backends-infra (no external package required)
- Feat: Add desktop notification support for AI response completion (experimental), suggested and reviewed by Silex
- Fix: Couple of bugs inside ai-code-prompt-mode.el, reported and fix suggested by lesteral
- Chore: C-u on ai-code-cli-start will prompt user to enter CLI args
- Feat: Better support for task management
- Fix: ai-code–setup-snippets silently fails, reported by lesteral
- Fix: add back multiple session support
- Feat: add kiro-cli support, by jenkinsjb
- Feat: add cursor-cli support, by donneyluck
- Fix: Weird ai-code-backends-infra.el loading bug, by Silex, and opencode start issue, by e665107
- Feat: Integrate Gemini CLI into ai-code-backends-infra, removing dependency on external gemini-cli.el package.
- Feat: Add ai-code-backends-infra.el, use it as infrastructure for
- OpenAI Codex
- GitHub Copilot CLI
- Opencode
- Grok CLI
- Feat: Add ai-code-build-project, let AI build the project
- Chore: Add AI-assisted test run fallback for running tests
- Chore: ai-code-take-notes prioritize to put note to visible org buffers in current frame
- Refactor ai-code-take-notes to smaller functions
- Feat: Support cross git repo context adding
- Chore: Add __pycache__ to default gitignore file patterns
- Fix: a couple of opencode related config / command
- Published to MELPA, update README and package description
- Feat: Add ai-code-insert-function-at-point, it is useful during writing TODO comment, to implement a feature referring to an existing function, from visible buffers
- Feat: Explain code in scope feature added choices: A. files visible; B. git repository
- Chore: Update AI backend upgrade & resume commands
- Fix: Improve the gptel generated shell command (C-c a !) should be one liner
- Fix: Fix errors and warnings reported by melpazoid workflow
- Feat: Add unit-test running support for TDD cycle
- Chore: Improve TDD flow, red and green stage
- Feat: Add Dired support for code change operation prompts
- Feat: Add Grok CLI backend support, by @richard134
- Feat: Add toggle buffer window dedication command
- Chore: Make ai-code-implement-todo flexible to modify code in the context. When it is a pure new code block, place it after the comment.
- Feat: Improve Implement TODO feature
- ai-code-implement-todo (C-a i) on blank line will ask user and insert TODO comment.
- After ai-code-implement-todo (C-a i) finishing code generation, mark TODO as DONE, instead of delete it
- ai-code-implement-todo support use clipboard content as context when C-u prefix is pressed
- ai-code-code-change (C-a c) on TODO comment line will use comment as code change instruction
- Feat: Add OpenCode backend support, requested and validated by gkzhb
- Feat: Introduce ai-code-sed-command, by sandinmyjoints
- Chore: Let AI to automatically determine the placeholder variable name during refactoring
- Chore: Improve ai-code-take-notes to prompt from open org buffers
- Fix duplicate .gitignore entries in ai-code-update-git-ignore
- Feat: Add context collection list management functions
- Feat: Add note taking functions, useful during discussing with AI
- Fix function detection for TODO comments preceding method definitions
- Addressing Wrong function detection, suggested by @Silex
- Refactor: error investigation prompt, run command with comint buffer
- Chore: Refactor error investigation prompts to improve context handling
- Chore: Add clipboard context option to ai-code-send-command function
- Chore: add compilation buffer content as error context
- Chore: Refactor AI shell command prompt to include marked files in dired mode
- Fix: work around claude-code-send-command signature change that broke programmatic command sending
- Feat: Add comprehensive refactoring techniques catalog and parameter
- Feat: MCP debug support and prioritize inspector.sh script
- Feat: add copilot resume via cli select ui, by shroomist
- Feat: add web browser support for git diffs
- Feat: generate python MCP config (json / toml)
- Chore: Update .gitignore with AI Code entries
- Feat: Add menu item to upgrade backend AI CLI tools
- Chore (inspector): migrate to eshell for improved command handling
- Chore: Enhance AI shell command input to support file name expansion in dired mode
- Feat: Add MCP debug command and customizable run options for Python files
- be able to insert ai generated command to shell buffer
- handle region and improve input
- Chore: Enhance ai-code-shell-cmd to
- Feat: add mcp inspector automatically runner
- Feat: add ai-code-init-project to init .projectile and create tags
- Feat: support run shell command / AI prompt shell command in dired buffer
- Chore: C-u on ai-code-implement-todo will keep the requirement
- Feat: add support for OpenAI Codex CLI backend
- Feat: add resume functionality for CLI sessions across backends, by ileixe