Laabam AILaabam AI
Laabam AILaabam AI
New chatBots

Sign up to save your chat history.

See plans and pricingSettingsHelp

Terms · Privacy · Refunds · Contact

Get more with a free account

Log in to save your chats, organise them in projects, attach PDFs and Word documents, and ask the Laabam Bots.

Log in
  1. Home
  2. Laabam AI Coder
Laabam AI

Laabam AI Coder

A coding assistant for VS Code, powered by Laabam Agni. Describe what you want in plain words: it reads your project, plans the change, edits files and runs commands, and you approve each step.

Coming soon to the VS Code MarketplaceGet an API key
Plan and Act
Talk a change through in Plan mode, then switch to Act mode to have it made.
Edits you approve
Every file change opens as a diff you can accept, edit or reject.
Runs your commands
Builds, tests and dev servers run in your terminal, and it reads their output.
Checkpoints
Each step is saved, so you can compare it with your workspace or roll back.
Rules and workflows
Project instructions and repeatable steps, kept in your repository.
MCP servers
Connect Model Context Protocol servers to give it more tools.

Get started

  1. Install Laabam AI Coder in VS Code and open it from the Laabam star in the activity bar.
  2. Make an API key at platform.laabam.ai and paste it in. It stays in VS Code's secret storage on your computer.
  3. Top up your credit at www.laabam.ai/credit.
  4. Type a task, such as "Add a dark mode toggle to the settings page".

Laabam Agni reads text only, so images can't be attached to a task.

Pricing

Calls are paid from your prepaid Laabam credit: ₹50 per 10 lakh tokens, the tokens Laabam AI Coder sends and the tokens of the answer alike, plus 18% GST in India. Top-ups start at ₹500 and are paid on www.laabam.ai/credit. Your balance and each call's tokens are on platform.laabam.ai.

Auto-approve

By default Laabam AI Coder asks before it reads, edits or runs anything. At the bottom of the Laabam AI Coder panel, click Auto-approve to turn it on, then the gear beside it to choose what it may do without asking: read project files, edit project files, run commands it judges safe (or all commands), or use MCP servers. There you can also set how many requests it may make in a row before it asks you again.

Everything it does still shows in the chat, and checkpoints let you roll back.

Rules

Rules are instructions Laabam AI Coder follows in every task, such as your coding style or how to run the tests. Put them in a .laabamrules file at the root of your project, or in Markdown files in a .laabamrules folder. Rules for all your projects go in Documents/Laabam AI Coder/Rules. Turn each rule file on or off from the rules button in the chat box.

To keep files out of its reach, list them in a .laabamignore file at the root of your project, written like a .gitignore.

Workflows

A workflow is a Markdown file of steps you repeat, such as preparing a release. Put workflows in .laabamrules/workflows in your project, or in Documents/Laabam AI Coder/Workflows for every project, and start one by typing / and its file name in the chat.

MCP servers

MCP servers give Laabam AI Coder more tools, such as a database or an issue tracker. Open MCP Servers from the server icon at the top of the Laabam AI Coder panel, then add a server: a remote one by its URL, or a local one by editing the MCP settings file it opens. The Model Context Protocol site lists servers and explains their settings.

Troubleshooting

The first answer takes a long time, or fails
Between 11 pm and 7:30 am IST, Laabam Agni starts up when the first request arrives, which can take up to about 15 minutes, and requests made meanwhile can fail after 5 minutes. Click Retry in a few minutes.
"401" or "invalid API key"
The key in Settings → Laabam Agni was deleted or mistyped. Make a new one at platform.laabam.ai and paste it in again.
"Your Laabam credit has run out"
Top up at www.laabam.ai/credit, then click Retry. Your balance is on platform.laabam.ai.
"Context window exceeded"
Laabam Agni reads up to 65,536 tokens at a time. When a task grows past that, Laabam AI Coder drops its oldest messages and tries again. For a long piece of work, start a new task for each part.
A file comes back with "rest of the code unchanged" or similar
An answer can be at most 16,384 tokens, so a very large file can be cut short. Ask for the change in smaller steps, or for one part of the file at a time.
"Shell integration unavailable"
Laabam AI Coder can't read the output of commands in that terminal. Update VS Code and use zsh, bash, fish or PowerShell as the default terminal profile (Command Palette → Terminal: Select Default Profile).
Checkpoints don't work
Checkpoints need Git installed on your computer. Install it from git-scm.com and restart VS Code.

Privacy

Your task, and the code and command output Laabam AI Coder reads to do it, go to laabam.ai to be answered by Laabam Agni, and to no other AI provider. laabam.ai records each call's token counts for your usage and bill. The extension itself sends no usage data. The Privacy Policy explains the rest.

Help

Type /reportbug in the chat to email a bug report to Laabam support, or write to support@laabam.ai.

Laabam AI Coder is a service of Laabam One Business Solutions Private Limited, built on Cline, which is used under the Apache License 2.0.

Terms of Service · Privacy Policy · Contact