caliber-ai-org/ai-setup

Continuously sync your AI setups with one command. Codebase tailor suited agent skills, MCPs and config files for Claude Code, Cursor, and Codex.

What it solves

Caliber prevents AI coding agents from using outdated project context. It solves the problem of hand-written context files (like CLAUDE.md or .cursor/rules/) becoming stale after code refactors, which leads to AI hallucinations and incorrect architectural advice.

How it works

Caliber automates the generation and maintenance of AI context files across multiple platforms. It analyzes a project's languages, frameworks, and architecture to create tailored configuration files. It uses a deterministic scoring system to audit existing configs by cross-referencing them against the actual filesystem to ensure paths and dependencies are accurate.

To keep context fresh, it employs pre-commit hooks and session-end triggers that run a refresh loop, updating the documentation based on recent git diffs. It also includes a "session learning" feature that captures tool usage and corrections during AI coding sessions to distill operational patterns into a learnings file.

Who it’s for

Developers and teams using AI coding agents such as Claude Code, Cursor, GitHub Copilot, OpenAI Codex, or OpenCode who want to ensure their AI assistants stay synchronized with their evolving codebase.

Highlights

  • Multi-Agent Support: Generates configs for Claude Code, Cursor, GitHub Copilot, Codex, and OpenCode simultaneously.
  • Deterministic Scoring: Audits config quality and grounding without requiring LLM calls.
  • Auto-Refresh: Uses git hooks to automatically update context files as code changes.
  • Session Learning: Learns from actual coding sessions to document patterns, gotchas, and fixes.
  • MCP Server Discovery: Automatically detects project tools and configures matching Model Context Protocol (MCP) servers.
  • Safe Workflow: Includes a review-first approach with diffs, automatic backups, and a full undo command.

Related

  • Project
    microsoft/agentrc

    A context engineering tool that automatically generates and maintains codebase-specific instructions and configurations to improve the performance of AI coding agents.

  • 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
    TokenRollAI/llmdoc

    A doc-driven workflow for Claude Code and Codex that automates the creation and synchronization of project documentation to maintain AI context.

  • Project
    JamesShi96/project-butler

    A persistent memory system for AI coding assistants that uses a structured Markdown-based memory stack to maintain project context, rules, and progress across sessions.

  • Project
    divar-ir/ai-doc-gen

    An AI-powered documentation generator that uses a multi-agent architecture to analyze codebases and automatically create READMEs and AI assistant configuration files.