moltext

Moltext

Note: Moltext was previously released as ContextMD.

Agent-native documentation compiler for Moltbots and autonomous agents.

Human documentation is optimized for humans, not agents. Agents ingest context; they do not navigate websites. Moltext is the compile step between human docs and agent memory. It output is a deterministic context.md.

Moltbots & Agent Systems

Moltbots and other autonomous agents require single-file, noise-free context to operate reliably. context.md is designed to be dropped directly into Moltbot workflows (memory or RAG pipelines). Moltext exists to standardize agent-readable documentation, ensuring deterministic input, memory stability, and entropy reduction.

Pipeline

• Parse: Crawl the documentation site.
• Normalize: Clean and standardizing the HTML.
• Compile: Use an LLM to structurally compress the content into agent-readable form.
• Emit: Generate the final context.md.

Installation

npm install -g moltext

Usage

moltext [options] <url>

Options

• -k, --key <key>: API Key (optional for local LLMs)
• -u, --base-url <url>: API Base URL (default: https://api.openai.com/v1)
• -m, --model <model>: Model name (default: gpt-4o-mini)
• -o, --output <path>: Output file path (default: context.md)
• -l, --limit <number>: Max pages to parse (default: 100)

Examples

Standard Usage (OpenAI):

moltext https://docs.example.com -o context.md

Using a Moltbot / Local Model (Ollama): Instruct the compiler to use the same brain as your agent.

moltext https://docs.example.com \
  --base-url http://localhost:11434/v1 \
  --model llama3

License

© Udit Akhouri — Moltext

CC-BY-NC-4.0