TeleCode AI
Remote-controlled autonomous coding agent in your VS Code.
Run tasks locally or from Telegram/WhatsApp, with grouped logs, status phases, and a focused coding toolchain.
TeleCode AI is an autonomous coding agent built on @mariozechner/pi-agent-core and @mariozechner/pi-ai.
It plans, executes tools, edits files, and reports progress in a structured log UI.
You can control it from VS Code or remotely via Telegram/WhatsApp.
🧱 Architecture
src/extension.ts is intentionally thin and acts as the VS Code composition root.- Runtime, channels, command wiring, settings sync, fetch logging, and UI status orchestration live in focused modules under
src/extension/.
- Agent tools are defined in
src/tools/index.ts and src/tools/definitions/*.
- Telegram and WhatsApp channels share common runtime bootstrap logic, while channel-specific rendering and message helpers live next to each channel.
🚀 Features
- Autonomous Agent Loop: Continuously plans and executes multi-step tasks. The agent reads your workspace, runs bash commands, edits files, and loops until the task is done.
- Remote Control Channels: Use Telegram and WhatsApp channels for remote task execution from your phone.
- Multi-Provider AI: Works with OpenRouter, OpenAI, Anthropic Claude, Google Gemini, MiniMax, Moonshot (Kimi), and self-hosted Ollama models.
- Structured Logs UI:
- Grouped log view with typed rows (
status, tool:start, tool:done, channel, llm, etc.)
- Tool invocations are visible in the task timeline, including remote channel runs
- Status verbosity modes:
minimal, normal, debug
- Filter chips and query filter for fast navigation in long runs
- Collapse/expand all grouped nodes
- Native Coding Toolchain:
read_file / write_file / edit_file (exact-line replace)fetch_url (read readable text from any HTTP website without external APIs)diagnostics (instantly read VS Code TypeScript/ESLint errors without running compilers)glob for file discoverygrep (ripgrep with Node.js fallback)bash with captured stdout/stderr, timeout, and streamingworkspace context: active file, cursor, open documents
- Configurable Response Style:
concise, normal, detailed.
- Session History: Persistent session memory within a VS Code workspace.
💡 Why a VS Code Extension?
- Zero Context Switching: TeleCode AI lives directly in your sidebar. No terminal juggling.
- Native Context: The agent instantly knows your Active File, Cursor Position, and Selected Text.
- Instant Feedback: The agent modifies the exact files you have open. See live diffs without jumping between apps.
- Remote Control via Telegram:
- 🚑 Emergency fix on the go: A bug is reported. Message the bot from your phone — it fixes and commits.
- 🛏 Work from bed: Your dev environment is on your office PC. Just leave VS Code open and instruct remotely.
- 🚶 Background tasks: Ask the agent to build a feature, go for a walk. Code is waiting when you return.
🛠 Getting Started
- Install from the VS Code Marketplace
- Open the TeleCode AI panel from the Activity Bar
- Configure your provider and API key in Settings (
Cmd/Ctrl + , → TeleCode AI)
- Click Start TeleCode, type a task in the prompt field, and send it
📱 Remote Commands
TeleCode AI is designed to be useful from messengers first. Telegram and WhatsApp share the same practical remote flows:
run/review/checks for daily task executionqueue/history/task/cancel for task controlcommit/revert/artifacts for result handoffmemory/remember/forget for workspace-scoped project memorygit/changes/diff/logs for lightweight remote inspectionschedule for periodic remote tasks
Typical examples:
/run fix failing tests in taskRunner and show me the result
/review
/history 10 failed
/task last
/artifacts last
/memory
/remember repo uses pnpm only
/git status
/schedule every 1 send me the latest git commit title
The VS Code panel remains useful for logs and local runs, but the main operational loop can be fully driven from Telegram or WhatsApp.
⚙️ Development
npm install # Install dependencies
npm run build # Build extension
npm test # Run test suite
npm run test:coverage # Run tests with coverage report
npm run watch # Watch mode
For hot reload: open Run and Debug → start Run Extension (Watch).
⚙️ Settings
| Setting |
Description |
telecode.provider |
AI provider (openrouter, openai, anthropic, google, moonshot, ollama, ...) |
telecode.model |
Model ID (e.g. openai/gpt-4o, arcee-ai/trinity-large-preview:free) |
telecode.apiKey |
API key for the selected provider |
telecode.baseUrl |
Custom base URL for OpenAI-compatible endpoints |
telecode.maxSteps |
Max agent steps per task (default: 100) |
telecode.responseStyle |
Response style (concise, normal, detailed) |
telecode.language |
Agent language (auto, ru, en) |
telecode.uiLanguage |
UI language (ru, en) |
telecode.statusVerbosity |
Runtime status density in logs (minimal, normal, debug) |
telecode.safeModeProfile |
Permission profile (strict, balanced, power) |
telecode.logMaxChars |
Maximum in-memory UI log buffer before trimming |
telecode.channelLogLines |
Maximum per-channel log lines kept in memory |
telecode.allowOutOfWorkspace |
Allow agent file/command access outside workspace |
telecode.allowedTools |
Allowed tool list for the agent |
telecode.telegram.enabled |
Enable Telegram bot |
telecode.telegram.botToken |
Token from @BotFather |
telecode.telegram.chatId |
Allowed Telegram user/group ID |
telecode.telegram.apiRoot |
Telegram API root URL |
telecode.telegram.forceIPv4 |
Force IPv4 mode for Telegram connectivity |
telecode.whatsapp.enabled |
Enable WhatsApp channel |
telecode.whatsapp.sessionPath |
Local session/profile path for WhatsApp Web auth |
telecode.whatsapp.allowSelfCommands |
Allow self-chat command mode (/run, /status, /stop, /help) |
telecode.whatsapp.accessMode |
WhatsApp sender policy (self, allowlist, all) |
telecode.whatsapp.allowedPhones |
Comma-separated allowlist phones for allowlist mode |
⏱ Remote Scheduler
You can create recurring remote tasks directly from Telegram or WhatsApp:
/schedule every 1 send me the current date and time
/schedule every 5 send me the latest git commit title
/schedule
/schedule run 1
/schedule pause 1
/schedule resume 1
/schedule remove 1
Notes:
- Minimum interval is
1 minute.
- Schedules are persisted in the workspace under
.telecode/remote-schedules.json.
- Scheduled runs reuse the same task queue, so they do not bypass active manual tasks.
📄 License
MIT © Uginy
| |
