Configuration

All configuration lives in .instar/config.json, created during setup and editable at any time.

Server

{
  "server": {
    "port": 4040,
    "host": "127.0.0.1"
  }
}

Field	Default	Description
`port`	4040	Server port
`host`	127.0.0.1	Bind address (localhost only by default)

{
  "telegram": {
    "botToken": "...",
    "chatId": -100...,
    "topicAutoCreate": true
  }
}

{
  "whatsapp": {
    "enabled": true
  }
}

WhatsApp state is managed internally via the Baileys library.

Scheduler

{
  "scheduler": {
    "enabled": true,
    "maxConcurrentSessions": 3
  }
}

Auth

{
  "auth": {
    "token": "..."
  }
}

The auth token is used for API authentication. Generated during setup.

Serendipity Protocol

{
  "serendipity": {
    "enabled": true,
    "maxPerSession": 5
  }
}

Field	Default	Description
`enabled`	`true`	Enable/disable the serendipity capture protocol
`maxPerSession`	`5`	Maximum findings a sub-agent can capture per session

The protocol is opt-out — enabled by default. Findings are stored in .instar/state/serendipity/.

Identity Files

These aren’t in config.json but are critical configuration:

File	Purpose
`.instar/AGENT.md`	Agent identity — who it is, its principles
`.instar/USER.md`	User context — who it works with, preferences
`.instar/MEMORY.md`	Persistent learnings across sessions
`.instar/ORG-INTENT.md`	Organizational constraints (optional)

Jobs

Jobs are defined in .instar/jobs.json. See Job Scheduler for the format.

Hooks

Behavioral hooks are installed in .claude/settings.json and scripts live in .instar/hooks/ and .claude/scripts/. See Hooks reference for details.