Chetna
Your codebase, understood.
An AI coding assistant for VS Code that indexes your workspace, retrieves what matters, and applies changes with review you can trust.
Context
Only the code that matters.
A four-layer symbol index, semantic retrieval, and a live call graph keep Chetna grounded in your project — not a generic guess.
- Layered index Files → classes → functions → details, kept in sync as you work.
- Codebase RAG Retrieves the slices that answer the question, within your context budget.
- Code graph Query paths, impact, and explanations from resolved calls and imports.
Edits
Change with a paper trail.
Guided generation, permission prompts, and a collapsible diff panel — review, keep, or revert per file before anything sticks.
Models
Local first. Cloud when you want.
One-click local setup with a starter model, or OpenAI, Claude, and Gemini — switch from the composer without leaving the editor.
Workflow
From question to change.
-
01
Index
Chetna maps your workspace and builds retrieval + graph layers.
-
02
Ask
Chat with active file, git diff, and context size at your fingertips.
-
03
Review
Approve tools, inspect diffs, keep what you want.
Install
Open Chetna in five minutes.
-
Marketplace
Search Chetna Editor by DistractionsForge, or install
DistractionsForge.chetna-editor. -
Model
Run
Chetna: Set Up Local Model, or add a cloud API key in settings. -
Chat
Cmd/Ctrl+Shift+C — index the workspace, then ask.
Current release
0.5.2
Local model polish, code graph tools, and a quieter path from install to first useful answer.
- Graph query, path, explain, and affected — built in-process from your index
- Unified local model setup and choose flow
- Recommended starter: Gemma 4 E4B with 64k context