Reference

Technical documentation covering how Manicode prompts work, their security model, model-specific behavior, and machine-readable output formats. This section is for deeper understanding — you don't need it to get started.

🗃️ Architecture and Design

1 item

🗃️ Security Assumptions

1 item

🗃️ Model-Specific Guidance

1 item

🗃️ Output Schemas

1 item

Section Overview

Architecture and Design

How prompts are structured, trust boundaries between components, and the security invariants that Manicode maintains. Start here if you want to understand the design principles behind the prompt library.

Security Assumptions

What Manicode guarantees and what it does not. Documents the threat model, the boundaries of LLM-based security tooling, and the assumptions you should validate in your own environment.

Model-Specific Guidance

Choosing between supported models (Claude Opus 4.6, GPT 5.3 Codex, Gemini 3.1 Pro, Grok 4.1, GitHub Copilot), behavioral differences across models, determinism considerations, and cost tradeoffs.

Output Schemas

Machine-readable output formats for integrating Manicode outputs into automated pipelines and reporting tools.