praisonai-bot Package Migration

Where bots, gateway, and channel CLI live after the C9 extraction from the praisonai wrapper into a new sibling PyPI package

praisonai-bot is a new sibling PyPI package holding the bots, gateway, and channel CLI — extracted from the praisonai wrapper — and every existing import still works via alias_package shims.

from praisonaiagents import Agent

agent = Agent(
    name="Support",
    instructions="Answer support questions from users on Telegram, Slack, and Discord.",
    llm="gpt-4o-mini",
)
agent.start("Hello — reply if you see this.")

A gateway process running praisonai-bot gateway start routes each Telegram/Slack/Discord message to this agent and posts the reply back.

Quick Start

1. Nothing changes if you install the wrapper

   pip install praisonai pulls in praisonai-bot transitively. Every existing command still works:

   pip install praisonai
   praisonai bot serve --config bot.yaml
   praisonai serve gateway --config gateway.yaml

2. Standalone install for headless workers or containers

   Install only what your container needs:

   pip install praisonai-bot[gateway,bot]
   praisonai-bot gateway start --config gateway.yaml
   praisonai-bot bot start --config bot.yaml

3. Agent-centric example — route messages from any channel

   from praisonaiagents import Agent
   
   agent = Agent(
       name="Support",
       instructions="Answer support questions from users on Telegram, Slack, and Discord.",
       llm="gpt-4o-mini",
   )
   agent.start("Hello — reply if you see this.")

   Run the gateway alongside it:

   praisonai-bot gateway start --agents agents.yaml

   Messages arrive from any configured channel, the agent replies, and the gateway posts the response back.

How It Works

The gateway is the only process that holds channel credentials. Agents never talk directly to Telegram, Slack, or Discord — they receive plain text and return plain text. The gateway handles platform-specific formatting, rate-limiting, and delivery guarantees.

Which install shape is right for me?

Compatibility Guarantees

Every existing import path continues to resolve. No code changes required.

• pip install praisonai remains the canonical user-facing install.
• praisonai.bots., praisonai.gateway., praisonai.daemon. are stable API — they are alias_package shims backed by praisonai_bot..
• Shim registration is safe under concurrent module imports — if another thread mutates sys.modules while shims are being installed (e.g. an async client loading modules during a parallel test run), the shim retries the snapshot and falls back to key-by-key reads instead of raising RuntimeError: dictionary changed size during iteration.
• praisonai-bot has no hard PyPI dependency on praisonai — no import cycle.
• Any access that praisonai-bot needs into the wrapper goes through _wrapper_bridge at call time, degrading gracefully when the wrapper is absent.

New CLI: praisonai-bot

The praisonai-bot binary is separate from the praisonai binary. It exposes:

praisonai-bot
├── bot         Start messaging bots with full agent capabilities
│   ├── start       Start a bot from a YAML config file (zero-code)
│   ├── telegram    Start a Telegram bot
│   ├── discord     Start a Discord bot
│   ├── slack       Start a Slack bot
│   └── ...
├── gateway     Manage the PraisonAI gateway server
│   ├── start       Start the gateway server
│   ├── stop        Stop a running gateway instance
│   └── status      Check gateway and daemon status
├── pairing     Manage bot pairing
├── identity    Manage bot identity
├── onboard     Onboard a new bot
├── kanban      Kanban integration
├── claw        Channel-agnostic bot operations
└── mint-link   Mint magic links

Examples:

praisonai-bot gateway start --config gateway.yaml
praisonai-bot gateway start --agents agents.yaml --port 9000
praisonai-bot gateway stop
praisonai-bot gateway status

praisonai-bot bot start --config bot.yaml
praisonai-bot bot telegram --token $TELEGRAM_BOT_TOKEN --model gpt-4o-mini
praisonai-bot bot slack --token $SLACK_BOT_TOKEN --memory

praisonai-bot mint-link                 # hyphen form (matches --help output)
praisonai-bot mint-link --ttl 600       # with custom TTL
praisonai-bot mint_link                 # underscore form (backward-compatible)

Extending: Register a New Channel

praisonai-bot discovers channel adapters via the praisonai.channels entry-point group at startup. No changes to praisonai-bot itself are needed.

To publish a third-party channel adapter, add an entry-point in your package’s pyproject.toml:

[project.entry-points."praisonai.channels"]
mattermost = "my_pkg.mattermost:MattermostBot"

The gateway/bot loader calls entry_points(group="praisonai.channels") on startup and registers every discovered adapter automatically.

Built-in Channels

Behavior changes

Recent fixes to the praisonai bot ... CLI that operators upgrading should know about:

Best Practices

Related