AI Capabilities

Zcrafter leverages Large Language Models (LLMs) to act as an intelligent pair programmer for mainframe development.

AI Agents

Zcrafter uses specialized "Agents" for different tasks:

Coder Agent: The primary agent for answering questions, writing code, and analyzing problems. It has access to your current file context and project structure.
Summarizer Agent: Generates concise summaries of long outputs or file contents.
Task Agent: Breaks down complex requests into smaller, actionable steps.
Title Agent: Generates short, descriptive titles for your conversation history.

The AI is only as good as the context it has. Zcrafter automatically manages context for you:

Current File: When you have a file open (e.g., a COBOL program), its content is sent to the AI.
Conversation History: The AI remembers previous questions and answers in the current session.
Project Context: You can configure the AI to be aware of your entire project structure (see Configuration).

To get the best results for COBOL, JCL, and PL/I:

Be Specific: Instead of "Fix this," say "Fix the S0C7 abend in the calculation paragraph."
Provide Error Codes: Always include ABEND codes (e.g., S0C4, S806) or error messages.
Ask for Explanations: "Explain what this JCL step does line-by-line."

Zcrafter supports multiple AI providers to give you flexibility in cost and performance:

OpenRouter (Recommended): Access to top models like Claude 3.5 Sonnet, GPT-4o, and Llama 3.
Anthropic: Direct access to Claude models.
OpenAI: Direct access to GPT-4 and GPT-3.5.
Google Gemini: Access to Gemini Pro and Flash models.
Groq: Ultra-fast inference for open-source models.
xAI: Access to Grok models.

OpenRouter

Anthropic

OpenAI

Google Gemini

Groq