Al Hakeem is an AI-powered coding assistant built directly into Visual Studio [2022, 2026], designed to help developers write, understand, and improve code faster. It combines intelligent autocomplete, conversational chat, agentic tool calling, and deep “thinking” capabilities...
AI-powered coding assistant using local LLM server. Features: Chat, Code Edit, Apply, Autocomplete, and Embeddings. You will find it in the "View" top menu after installation. Tested with Lemonade, LM Studio and Ollama. Recommendation use with Lemonade.
Features:
Code Assistant.
Autocomplete.
Agentic Coder.
Supports OpenAI API [Lemonade, LM Studio, Ollama]
Roadmap:
MCP
Code review before apply
Drag and Drop files (allow images for vision models)
List of file changes
Add Embedding files (pdf, doc, docx)
Add references to files using @
Run multiple Agents?
Run in Debug mode?
Menu Context actions similar to Copilot
TBD
Release notes:
v0.1.41:
Fixed an issue with get_file where it reports the wrong number of lines.
Fixed an issue with finish_plan file naming.
Fixed an issue with Summarize by removing writing to file.
v0.1.40:
Added a checkbox for .BAK files in the Settings>General tab.
Added bubble timer.
Added extra cases handling for get_file and replace_string.
v0.1.39:
Added tokens stats for assistant messages.
Fixed some debug logs.
v0.1.38:
Fixed an issue where that chat still thinks it is still going after an exception (UI glitch).
Added renaming PLAN.md according to plan subject upon plan finish.
v0.1.37:
Fixed an exception issue with run_build
v0.1.36:
Fixed an issue with run_build where it reports build success when an exception happens.
v0.1.35:
Changed OpenAI API to Llama.cpp API
Removed Current file from prompt to lower input tokens count
Added max results for grep_search, file_search, code_search, get_tests
Added ignore build warnings for run_build in the settings menu
v0.1.34:
Added scroll to end of chat button
v0.1.33:
Fixed an issue with scrolling up and down inside the chat window not working sometimes.
v0.1.32:
Fixed an issue with get_tests tool call.
Added OpenAI API
v0.1.31:
Updated chat format to Markdown
v0.1.30:
Fixed an issue with file_search when the search query has a path.
Fixed multiple button clicks issues that might cause VS to crash.
Added deselecting code after sending.
Added proper exception handling for tool callings.
v0.1.29:
Fixed thinking control for Ollama
v0.1.28:
Added Provider selection
Added resizable chat area
v0.1.27:
Added Debug console
Added support for API Key
Added support for Unsloth Studio
Fixed an issue where thought text do not appear during chat bubble filling
v0.1.26:
Tweaked thinking prompts.
v0.1.25:
Added Dark theme checkboxes
Added coding conventions flags in Settings --> General Tab
v0.1.24:
Fixed UI glitch where chat bubble does not show thoughts has been received.
Cleaned up some empty messages due to connection issues.
v0.1.23:
Fixed chat backup when shrinking.
Fixed an issue when rejecting a question.
Added thoughts toggle to show or hide the thoughts in the chat bubble.
Fixed an issue chatting with LM Studio.
v0.1.22:
Added settings menu [Only for LLM and Tools]
Enhanced cutoff detection.
v0.1.21:
Fixed an issue where the tokens status not updating.
Fixed an issue where the get_web_pages not executing.
Fixed iteration counter where it shows wrong maximum iteration number right after answering a question.
v0.1.20:
Fixed an issue where the Max Output Tokens size selection does not go above 16K.
Fixed an issue where the DONE message is not picked up to signal an end of message or cutoff.
v0.1.19:
Added XXLong workload option for max_completion_tokens which uses 65536 tokens for output tokens.
Added Half workload option for max_completion_tokens which uses half of the full num_ctx of the selected model for output tokens.
Added MAX workload option for max_completion_tokens which uses the full num_ctx of the selected model for output tokens.
Added some text labels.
Added cutoff detection.
Added some information messages when cutoff occurs.
Enhanced support for LM Studio, Lemonade and Ollama.
v0.1.18:
Fixed an issue where run_build reports Build Failed when build succeeds but with warnings.
Clear chat will reset the 📝 Auto Commit Checkbox to false.
Added 🤏 Auto Shrink to automatically shrink conversation when it reaches to critical level of 97% or above.
Clear chat will reset the 🤏 Auto Shrink Checkbox to false.
Set http client timeout to infinite.
Added duration time for cancelled tasks.
Added separate URL and Port for autocomplete model [modify $AppData/.alhakeem/alhakeem.settings.json].
v0.1.17:
Fixed an issue where tokens usage bar does not go yellow when above 60%.
Fixed an issue where the settings gets overwritten.
Made settings global (AppData.Roaming).
Added showing version tag on title.
Added showing timestamp for loaded messages.
Fixed an issue where loaded chat messages do not look correct.
Fixed an issue where not all answers are sent back to LLM.
Fixed an issue where chat area scrolls up when losing then regaining focus.
v0.1.16:
Fixed ask_question broken json.
v0.1.15:
Fixed an issue where the loop skips the question when always approve is enabled.
v0.1.14:
Fixed an issue where at the start of the loop it skips the question when always approve is enabled.
Fixed an issue where questions are displayed on top of each other on first time display
v0.1.13:
Added chat timer
Fixed get_file tool call
Code cleaning by removing unused methods.
Chat backup when clearing chat.
v0.1.12:
Skipped.
v0.1.11:
Fixed grep_search
Enhanced file searching for files related tool calls
v0.1.10:
Fixed get_projects_in_solution
Fixed some label typos
Fixed chat window scrolling to top when sending
Fixed issue where chat scrolls up when switching tabs. now it scrolls down.
Added Auto Commit checkbox
Enhanced Autocomplete [still under work]
v0.1.9:
Added Thinking controls vis combobox [off, low, medium, high, xhigh].
Added reading max tokens from loaded model and reflect it on the token usage bar.
Added some tool tips.
Modified some tool tips.
Fixed inf loop detection.
Fixed an issue where csproj sometimes is not detected.
v.0.1.8:
Added proper plan handling (updating PLAN.md with progress bar).
Added opening PLAN.md file after each modification.
Fixed TPS. now it shows values from the LLM.
Added TTFS (Time to first token) and other stats.
v.0.1.7:
UI: Add model loading button, token bar, and refactor:
Added "Load Model" button to UI with context length support [not working yet]
Introduced ProgressBar for token usage with tooltips
Improved input field padding for better appearance
Fixed busy status emoji formatting for query investigation
Fixed TPS
v.0.1.6:
Refactor UI, file ops, and tool runners; bump to 0.1.6
Added auto git commit.
Refactored using directives for consistency and removed unused imports.
Improved directory/file filtering to exclude build and hidden folders.
Enhanced chat UI with assistant thoughts display, centralized role tags, better button state management, session ID tracking.
Improved model load/unload logic.
Added Infinite loop protection.
Updated tool runners for clearer messages.
Improved file edit service with git checkpoint helper and better commit messages.
Updated model unloading and streaming.
Refactored test discovery/execution for reliability.
