Configuration
All configuration is via environment variables. Copy .env.example to .env and fill in the values.
Required
# Slack (get these from api.slack.com/apps)
SLACK_BOT_TOKEN=xoxb-... # Bot User OAuth Token
SLACK_APP_TOKEN=xapp-... # App-Level Token for Socket Mode
# LLM (at least one)
ANTHROPIC_API_KEY=sk-ant-... # For Claude models
See SLACK_SETUP.md for how to get the Slack tokens.
Optional
# GitHub (for repo operations)
GITHUB_TOKEN=ghp_... # Personal Access Token with 'repo' scope
# Alternative LLM providers
OPENAI_API_KEY=sk-... # For GPT models
OPENROUTER_API_KEY=sk-or-... # For other models via OpenRouter
# Model settings
DEFAULT_MODEL=gpt-5.2 # Default model to use
# Logging
LOG_LEVEL=info # debug, info, warn, error
LOG_DIR=./logs # Where to write log files
LOG_FORMAT=text # text or json
# Storage
USE_LOCAL_STORAGE=true # Uses ./data directory
LOCAL_DATA_DIR=./data # Where to store sessions/jobs/repos
# Agent limits
MAX_AGENT_ITERATIONS=50 # Max tool calls per task
WORKER_MAX_RUNTIME_MS=1800000 # 30 min timeout
MAX_CONCURRENT_JOBS=5 # Parallel jobs
Access Control
# Restrict usage (comma-separated Slack user IDs)
ALLOWED_USERS=U12345678,U87654321
# Restrict to specific channels
ALLOWED_CHANNELS=C12345678,C87654321
# Restrict which repos can be accessed
ALLOWED_REPOS=myorg/frontend,myorg/backend
BLOCKED_REPOS=myorg/secrets
Models
Claire supports multiple providers:
Anthropic (via ANTHROPIC_API_KEY):
claude-sonnet-4-5-20250929- Default, balanced (Claude 4.5)claude-opus-4-5-20250929- Most capable (Claude 4.5)claude-haiku-4-5-20250929- Fastest (Claude 4.5)
OpenAI (via OPENAI_API_KEY):
gpt-5.2- Default if setgpt-4.1
OpenRouter (via OPENROUTER_API_KEY):
- Any model available on OpenRouter
Users can switch models per-message with model=claude-opus-4-5-20250929 in their Slack message.
Example .env
# Required
SLACK_BOT_TOKEN=xoxb-1234567890-1234567890123-abcdefghijklmnop
SLACK_APP_TOKEN=xapp-1-A1234567890-1234567890123-abcdef
ANTHROPIC_API_KEY=sk-ant-api03-xxxx
# GitHub (optional but recommended)
GITHUB_TOKEN=ghp_xxxxxxxxxxxxxxxxxxxx
# Production settings
LOG_LEVEL=info
NODE_ENV=production