Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.charmos.io/llms.txt

Use this file to discover all available pages before exploring further.

Charm runs agents inside managed container images. Choose the smallest image that supports your workload.

​
Configure Runtime Mode

runtime:
  mode: "standard"
Valid values:
  • standard
  • full

​
Standard Runtime

Use standard for most Python agents. Best for:
  • text processing,
  • API calls,
  • lightweight data workflows,
  • RAG-style agents,
  • simple LangChain/LangGraph/custom agents.
It is optimized for lower cold-start time and smaller dependency footprint.

​
Full Runtime

Use full when the agent needs browser or multimedia support. Best for:
  • Playwright, Selenium, Puppeteer,
  • FFmpeg or audio/video workflows,
  • OpenClaw agents,
  • browser-rendered output.

​
Choosing a Mode

Start with standard. Move to full only when a dependency requires it.