* fix: recover from active runs stuck on approval recoverOrphanedConversationApproval() only resolved approvals from terminated (failed/cancelled) or abandoned (completed/requires_approval) runs. Active runs with status=running and stop_reason=requires_approval were skipped, even though they block the entire conversation. This happens when the CLI's HITL approval flow leaves a pending approval that no client will resolve (e.g. agent triggered via ADE, user walked away). Every subsequent message fails with success=false. Now also handles running+requires_approval: rejects the approval and cancels the stuck run so lettabot can proceed. Fixes the recovery path for all three call sites: pre-send check, CONFLICT catch, and error-result retry. Written by Cameron ◯ Letta Code "The best error message is the one that never shows up." -- Thomas Fuchs * fix: use cancelRuns return value for accurate diagnostics Check boolean return from cancelRuns() instead of always logging success. Details string now shows "(cancel failed)" when it didn't work. Written by Cameron ◯ Letta Code "Trust, but verify." -- Ronald Reagan * fix: detect terminal errors and retry with recovery in bot.ts Monkey patch until SDK issue letta-code-sdk#31 is resolved. 1. Detect terminal errors (success=false OR error field), not just empty-success. Both trigger orphan recovery + retry. 2. Blind retry on terminal error when no orphaned approvals found -- client-side approval failures may leave no detectable record. 3. sendToAgent() throws on terminal error instead of returning empty string. Background callers get actionable errors. Written by Cameron ◯ Letta Code "A good patch is one you can remove." -- unknown * test: add tests for approval recovery including stuck runs 7 tests covering recoverOrphanedConversationApproval(): - Empty conversation - No unresolved approvals - Recovery from failed run - Recovery from stuck running+requires_approval (with cancel) - Already-resolved approvals skipped - Healthy running run not touched - Cancel failure reported accurately Written by Cameron ◯ Letta Code "Code without tests is broken by design." -- Jacob Kaplan-Moss * fix: gate retry on sentAnyMessage to prevent duplicate delivery Codex review caught that finalizeMessage() clears the response buffer on type changes, so hasResponse could be false even when output was already sent. This caused duplicate retries with potential side effects. Now checks !sentAnyMessage (authoritative delivery flag) in addition to !hasResponse before retrying. Written by Cameron ◯ Letta Code "Idempotency: the art of doing nothing twice." -- unknown
LettaBot
Your personal AI assistant that remembers everything across Telegram, Slack, Discord, WhatsApp, and Signal. Powered by the Letta Code SDK.
Features
- Multi-Channel - Chat seamlessly across Telegram, Slack, Discord, WhatsApp, and Signal
- Unified Memory - Single agent remembers everything from all channels
- Persistent Memory - Agent remembers conversations across sessions (days/weeks/months)
- Local Tool Execution - Agent can read files, search code, run commands on your machine
- Voice Messages - Automatic transcription via OpenAI Whisper
- Heartbeat - Periodic check-ins where the agent reviews tasks
- Scheduling - Agent can create one-off reminders and recurring tasks
- Streaming Responses - Real-time message updates as the agent thinks
Quick Start
Prerequisites
- Node.js 20+
- A Letta API key from app.letta.com (or a running Letta Docker server)
- A Telegram bot token from @BotFather
Install
# Clone the repository
git clone https://github.com/letta-ai/lettabot.git
cd lettabot
# Install dependencies
npm install
# Build and link the CLI globally
npm run build
npm link
Update
# Pull latest changes
git pull origin main
# Reinstall dependencies and rebuild
npm install
npm run build
Optional: Run a Letta Docker server
You can use lettabot with a Docker server with:
docker run \
-v ~/.letta/.persist/pgdata:/var/lib/postgresql/data \
-p 8283:8283 \
-e OPENAI_API_KEY="your_openai_api_key" \
letta/letta:latest
See the documentation for more details on running with Docker.
Setup
Option 1: AI-Assisted Setup (Recommended)
Paste this into Letta Code, Claude Code, Cursor, or any AI coding assistant:
Clone https://github.com/letta-ai/lettabot, read the SKILL.md
for setup instructions, and help me configure Telegram.
You'll need:
- A Letta API key from app.letta.com (or a Letta Docker server)
- A Telegram bot token from @BotFather
The AI will handle cloning, installing, and configuration autonomously.
Option 2: Interactive Wizard
For manual step-by-step setup:
git clone https://github.com/letta-ai/lettabot.git
cd lettabot
npm install && npm run build && npm link
lettabot onboard
Run
lettabot server
That's it! Message your bot on Telegram.
Note: For detailed environment variable reference and multi-channel setup, see SKILL.md
Voice Messages
LettaBot can transcribe voice messages using OpenAI Whisper. Voice messages are automatically converted to text and sent to the agent with a [Voice message]: prefix.
Supported channels: Telegram, WhatsApp, Signal, Slack, Discord
Configuration
Add your OpenAI API key to lettabot.yaml:
transcription:
provider: openai
apiKey: sk-...
model: whisper-1 # optional, defaults to whisper-1
Or set via environment variable:
export OPENAI_API_KEY=sk-...
If no API key is configured, voice messages are silently ignored.
Skills
LettaBot is compatible with skills.sh and Clawdhub.
# from Clawdhub
npx molthub@latest install sonoscli
# from skills.sh
npm run skills:add supabase/agent-skills
# connect to LettaBot
lettabot skills
◆ Enable skills (space=toggle, enter=confirm):
│ ◻ ── ClawdHub Skills ── (~/clawd/skills)
│ ◻ 🦞 sonoscli
│ ◻ ── Vercel Skills ── (~/.agents/skills)
│ ◻ 🔼 supabase/agent-skills
│ ◻ ── Built-in Skills ──
│ ◻ 📦 1password
│ ◻ ...
# View LettaBot skills
lettabot skills status
Home Assistant
Control your smart home with LettaBot:
# 1. Install the skill from ClawdHub
npx clawdhub@latest install homeassistant
# 2. Enable the skill
lettabot skills sync
# Select "homeassistant" from the list
# 3. Configure credentials (see skill docs for details)
# You'll need: HA URL + Long-Lived Access Token
Then ask your bot things like:
- "Turn off the living room lights"
- "What's the temperature in the bedroom?"
- "Set the thermostat to 72"
CLI Commands
| Command | Description |
|---|---|
lettabot onboard |
Interactive setup wizard |
lettabot server |
Start the bot server |
lettabot configure |
View and edit configuration |
lettabot skills status |
Show enabled and available skills |
lettabot destroy |
Delete all local data and start fresh |
lettabot help |
Show help |
Channel Setup
LettaBot uses a single agent with a single conversation across all channels:
Telegram ──┐
Slack ─────┤
Discord ───┼──→ ONE AGENT ──→ ONE CONVERSATION
WhatsApp ──┤ (memory) (chat history)
Signal ────┘
- Start a conversation on Telegram
- Continue it on Slack
- Pick it up on WhatsApp
- The agent remembers everything!
| Channel | Guide | Requirements |
|---|---|---|
| Telegram | Setup Guide | Bot token from @BotFather |
| Slack | Setup Guide | Slack app with Socket Mode |
| Discord | Setup Guide | Discord bot + Message Content Intent |
| Setup Guide | Phone with WhatsApp | |
| Signal | Setup Guide | signal-cli + phone number |
At least one channel is required. Telegram is the easiest to start with.
Bot Commands
| Command | Description |
|---|---|
/start |
Welcome message and help |
/status |
Show current session info |
/heartbeat |
Manually trigger a heartbeat check-in |
Background Tasks (Heartbeats & Cron)
Heartbeats and cron jobs run in Silent Mode - the agent's text responses are NOT automatically sent to users during these background tasks. This is intentional: the agent decides when something is worth interrupting you for.
To send messages during silent mode, the agent must explicitly use the CLI:
lettabot-message send --text "Hey, I found something interesting!"
The agent sees a clear [SILENT MODE] banner when triggered by heartbeats/cron, along with instructions on how to use the CLI.
Requirements for background messaging:
- The Bash tool must be enabled for the agent to run the CLI
- A user must have messaged the bot at least once (to establish a delivery target)
If your agent isn't sending messages during heartbeats, check the ADE to see what the agent is doing and whether it's attempting to use lettabot-message.
Agent CLI Tools
LettaBot includes small CLIs the agent can invoke via Bash (or you can run directly):
lettabot-message send --text "Hello from a background task"
lettabot-react add --emoji :eyes: --channel discord --chat 123 --message 456
lettabot-history fetch --limit 25 --channel discord --chat 123456789
See CLI Tools for details and limitations.
Connect to Letta Code
Any LettaBot agent can also be directly chatted with through Letta Code. Use the /status command to find your agent_id, and run:
letta --agent <agent_id>
Security
Network Architecture
LettaBot uses outbound connections only - no public URL or gateway required:
| Channel | Connection Type | Exposed Ports |
|---|---|---|
| Telegram | Long-polling (outbound HTTP) | None |
| Slack | Socket Mode (outbound WebSocket) | None |
| Discord | Gateway (outbound WebSocket) | None |
| Outbound WebSocket via Baileys | None | |
| Signal | Local daemon on 127.0.0.1 | None |
Tool Execution
By default, the agent is restricted to read-only operations:
Read,Glob,Grep- File explorationweb_search- Internet queriesconversation_search- Search past messages
Access Control
LettaBot supports pairing-based access control. When TELEGRAM_DM_POLICY=pairing:
- Unauthorized users get a pairing code
- You approve codes via
lettabot pairing approve telegram <CODE> - Approved users can then chat with the bot
Development
# Run in development mode (auto-reload)
npm run dev
# Build for production
npm run build
# Start production server
lettabot server
Releases
Releases are automated via GitHub Actions. When a version tag is pushed, the workflow builds, tests, generates release notes from merged PRs, and creates a GitHub Release.
# Create a release
git tag v0.2.0
git push origin v0.2.0
# Create a pre-release (alpha/beta/rc are auto-detected)
git tag v0.2.0-alpha.1
git push origin v0.2.0-alpha.1
See all releases: GitHub Releases
Troubleshooting
Cannot find package 'keyv'
Error [ERR_MODULE_NOT_FOUND]: Cannot find package 'keyv'
Clean reinstall fixes this:
rm -rf node_modules package-lock.json && npm install
Session errors / "Bad MAC" messages These are normal Signal Protocol renegotiation messages. They're noisy but harmless.
Messages going to wrong chat Clear the session and re-link:
rm -rf ./data/whatsapp-session
lettabot server # Scan QR again
Signal
Port 8090 already in use
SIGNAL_HTTP_PORT=8091
General
Agent not responding Delete the agent store to create a fresh agent:
lettabot destroy
Heartbeat/cron messages not reaching my chat Heartbeats and cron jobs run in "Silent Mode" - the agent's text output is private and not auto-delivered. To send messages during background tasks, the agent must run:
lettabot-message send --text "Your message here"
Check the ADE to see if your agent is attempting to use this command. Common issues:
- Bash tool not enabled (agent can't run CLI commands)
- Agent doesn't understand it needs to use the CLI
- No delivery target set (user never messaged the bot first)
Documentation
- Getting Started
- Self-Hosted Setup - Run with your own Letta server
- Configuration Reference
- Slack Setup
- Discord Setup
- WhatsApp Setup
- Signal Setup
Acknowledgement
Some skills were adapted from Moltbot.
License
Apache-2.0