linghungegeg/Linghun

AGI-oriented, hallucination-resistant AI coding runtime grounded in evidence, tools, memory, agents, and verification.

What it solves

Linghun is an AI-powered programming terminal designed to move beyond simple chat-based suggestions. It addresses the common pitfalls of AI coding—such as hallucinations (claiming a task is done without verifying), model drift (forgetting project rules across sessions), and the instability of working in complex real-world environments like Windows with Chinese paths and PowerShell.

How it works

It acts as an "engineering exoskeleton" for LLMs. Instead of relying solely on prompts, Linghun implements system-level constraints. It integrates a provider runtime for multi-model routing, a codebase index for context awareness, and an evidence-first system that binds model conclusions to actual file reads, tool outputs, and Git states. It uses a "Workflow Matrix" to break complex goals into observable phases and roles, and a "Policy Kernel" to coordinate memory, permissions, and verification requirements before the model executes an action.

Who it’s for

  • Individual Developers: Those wanting to use natural language to drive real projects with built-in safety nets like Git stable points and automated verification.
  • Professional Engineers: Developers needing a structured AI workflow that handles codebase indexing, failure learning, and architectural boundary checks.
  • Enterprises: Teams requiring strict permission boundaries, local execution privacy, and the ability to connect internal software capabilities via a standardized App Bridge.

Highlights

  • Evidence-First System: Reduces hallucinations by requiring actual evidence (logs, file content, test results) before a task is marked as complete.
  • Windows-First Design: Native support for PowerShell, Chinese paths, and process guarding to ensure stability in Windows environments.
  • Architectural Guardrails: Includes "AntiCodeBlob" to prevent the AI from creating oversized "god files" or deep nesting.
  • Managed Workflows: Supports durable jobs, background tasks, and multi-agent coordination for long-running engineering tasks.
  • Extensible Ecosystem: Connects to external tools via MCP, Skills, and a Capability Runtime (App Bridge) for third-party software integration.

Related

  • Project
    Kiln-AI/Kiln

    A local-first AI development workbench that integrates prompt optimization, evaluations, RAG, and fine-tuning into a single workflow for teams.

  • Project
    NeoLabHQ/context-engineering-kit

    A plug‑in marketplace for LLM‑based coding assistants (Claude Code, Gemini, Antigravity, etc.) that supplies token‑efficient, context‑engineering patterns—reflection loops, sub‑agent judges, specification‑first development, code‑review agents, and Git helpers—to make AI‑generated code more reliable and easier to integrate into CI pipelines.

  • Project
    ExtensityAI/symbolicai

    A neuro-symbolic framework that integrates LLMs into Python by combining semantic reasoning primitives and a contract-based validation system to reduce hallucinations.

  • Project
    AgentEra/Agently

    An AI application runtime framework that provides structured outputs, observable actions, and signal-driven workflows to build reliable, production-ready AI service backends.

  • Project
    gszhangwei/open-spdd

    A methodology and CLI tool that transforms AI coding prompts into structured design contracts using the REASONS Canvas framework to ensure precise, traceable code generation.