agentlabs-porter

Overview

agentlabs-porter wraps the Qiscus AgentLabs pekabgt export/import API in two scripts, so Claude can back up, clone, migrate, or restore an AgentLabs bot config on request — “export this bot”, “back up the bot”, “import bot JSON into <code>”, “clone bot from staging”.

The token-hop chain (Omnichannel login → AgentLabs token exchange → export or import) is hidden inside the skill’s client library. The entry-point scripts emit a one-line JSON status envelope on stdout, so Claude can report success or failure without ever ingesting the bot JSON itself — which can be huge and may contain embeddings.

Capabilities

Both commands accept credential and base-URL overrides; the default AgentLabs base is https://api-chatbot.qiscus.com.

When to Use

• Snapshotting a bot before making risky changes
• Cloning a bot from staging into a prod-test environment
• Restoring a bot from a previous JSON backup

Not for: editing bot intents in the UI, running chatbot test scenarios, or general AgentLabs questions.

How to Access

1. Get the skill folder from the shared Drive folder
2. One-time setup:

cp agentlabs-porter/.env.example agentlabs-porter/.env
# fill OMNICHANNEL_EMAIL, OMNICHANNEL_PASSWORD
pip install requests python-dotenv

3. Export a bot:

python agentlabs-porter/scripts/export_bot.py --bot-code <BOT_CODE>

4. Import a bot:

python agentlabs-porter/scripts/import_bot.py --bot-code <BOT_CODE> --file <path-to.json>

Keep your .env out of version control — it holds your Omnichannel credentials. The skill’s repo ignores *.env by default.

Example Prompts

Export bot ABC123 and save the snapshot before I change its intents.

Clone the staging bot: export bot STG456, then import the JSON into PRD789.

Related Tools

• agentlabs-blueprint — design a new bot blueprint from requirements
• AgentLabs LLM Builder — generate importable AgentLabs JSON from a PRD