Using LLM Mode

Master natural language commands with the AI assistant.

Setup

1. Get an API key from platform.openai.com
2. Add to your .env:

OPENAI_API_KEY=sk-your-key-here
OPENAI_MODEL=gpt-4.1-nano

3. Restart the server

Basic Usage

Enable LLM Mode

In Telegram:

/llm

You’ll see:

LLM mode enabled! Send natural language commands.

Send a Request

Just type what you want to do:

show me the git status

The AI will:

1. Figure out the right command (git status)
2. Run it
3. Show you the output

Disable LLM Mode

/llm

Example Workflows

Git Operations

You: show me recent commits
AI: Running: git log --oneline -10
    [shows commits]

You: what changed in the last one?
AI: Running: git show HEAD --stat
    [shows changes]

You: create a branch called feature-auth
AI: Running: git checkout -b feature-auth
    Switched to a new branch 'feature-auth'

Code Exploration

You: find all config files
AI: Running: find . -name "*.config.*" -o -name "config.*"
    [shows files]

You: what's in the main config?
AI: Running: cat config/default.json
    [shows contents]

You: find where the API endpoint is defined
AI: Running: grep -r "api" --include="*.js" -l
    [shows files]

Project Management

You: run the tests
AI: Running: npm test
    [shows test output]

You: are there any failing tests?
AI: Based on the output, all 42 tests passed.

You: build the project
AI: Running: npm run build
    [shows build output]

File Operations

You: list files in the src folder
AI: Running: ls -la src/
    [shows files]

You: show me the first 20 lines of index.js
AI: Running: head -20 src/index.js
    [shows code]

You: find TODO comments
AI: Running: grep -rn "TODO" --include="*.js"
    [shows TODOs with line numbers]

Tips for Better Results

Be Specific

❌ "show files"
✅ "show all JavaScript files in the src directory"

Provide Context

❌ "what's wrong?"
✅ "the build failed, what's the error in the output?"

Ask Follow-ups

The AI remembers the conversation:

You: show me the package.json
AI: [shows file]

You: what version of express is installed?
AI: Based on the package.json, express version is 4.21.0

Conversation Management

Clear History

If the AI gets confused or you want to start fresh:

/llm_clear

Check Mode

Send /llm to see if LLM mode is enabled. It toggles on/off.

Model Selection

Choose a model based on your needs:

Set in .env:

OPENAI_MODEL=gpt-4o-mini

Limitations

• Commands timeout after 2 minutes
• Long outputs are truncated
• Max 10 commands per conversation turn
• Requires internet connection