MCP y llamadas a herramientas

AI Brain Starter

mycelium-hq/ai-brain-starter

An operating system and verification harness for Claude Code that provides persistent memory, accountability, journaling, and knowledge graph capabilities, ensuring context compounds across sessions rather than corrupting.

★ 33Estrellas
⑂ 25Bifurcaciones
53Problemas abiertos
PythonIdioma
MITLicencia
Q@project.QualityScorePuntuación editorial

Capturas de pantalla del proyecto

Captura de pantalla de AI Brain Starter Captura de pantalla de AI Brain Starter

Resumen

AI Brain Starter addresses the problem of LLM context corruption over multiple edit interactions by providing a deterministic verification harness around the AI agent. It uses a local markdown vault as ground truth instead of relying on chat history. The system enforces pre-write hooks to block bad edits, implements bi-temporal rule lineage for conflict and drift detection, and manages a strict session lifecycle to cascade decisions and insights to the correct files automatically.

Características principales

  • Vault as ground truth (Markdown on disk)
  • Deterministic pre-write hooks to block bad edits
  • Bi-temporal rule lineage and drift detection
  • Automated session lifecycle (context loading and end-of-session cascading)
  • Local-first architecture with no vector database or fine-tuning required
  • Idempotent bootstrap installer with backup and rollback capabilities
  • Corporate install profile for security reviews

Requisitos, instalación y guía rápida

macOS/Linux:
git clone https://github.com/mycelium-hq/ai-brain-starter ~/.claude/skills/ai-brain-starter 2>/dev/null || true
bash ~/.claude/skills/ai-brain-starter/bootstrap.sh

Windows (PowerShell):
if (-not (Test-Path "$env:USERPROFILE\.claude\skills\ai-brain-starter")) { git clone https://github.com/mycelium-hq/ai-brain-starter "$env:USERPROFILE\.claude\skills\ai-brain-starter" }
powershell -NoProfile -ExecutionPolicy Bypass -File "$env:USERPROFILE\.claude\skills\ai-brain-starter\bootstrap.ps1"

Alternatively, paste the setup prompt from the Install section into the Claude Code desktop app. To uninstall: bash bootstrap.sh --uninstall. To skip third-party vendor skills: set SKIP_VENDOR_SKILLS=1.

Uso

After installation, the system operates automatically within Claude Code sessions. It loads vault context at the start of every session. Users interact with their vault using commands like `/journal`. At the end of a session, the system automatically files decisions, insights, and to-dos into the appropriate vault files. Hooks operate transparently in the background to block invalid saves or edits.

Compatibilidad de modelos y casos de uso

Designed specifically as an operating system for Claude Code. Requires Claude Code 2.1.133 or newer and a paid Claude account (Pro, Max, or Team).

Notas sobre la licencia y los riesgos

The core ai-brain-starter repository is licensed under the MIT license. Installed third-party skill packs carry their own licenses (MIT, Apache 2.0, CC-BY-SA-4.0). One dependency (vercel-labs/agent-skills) has no license and is cloned for read-only reference.

Editorial verification 2026-08-02: repository URL, owner, description, license and repository statistics were reviewed. License metadata: MIT. README was fetched for the channel draft; re-check repository dependencies, releases and model terms before production use.

Lanzamiento y mantenimiento

Not stated in the repository metadata

MemPalace

mempalace/mempalace

★ 58KPython

QwenPaw

agentscope-ai/qwenpaw

★ 33,9KPython

Composio

composiohq/composio

★ 29,5KTypeScript