Back to browse
GitHub Repository

[MOVED] This repository has been moved to https://github.com/dance-of-tal/dance-of-tal

10 starsTypeScript

Dance of Tal – Decompose, mix, and reuse AI rules with an MCP server

by monarchjuno·Feb 24, 2026·2 points·0 comments

AI Analysis

●●SolidBig BrainShip It

Prompt composition via MCP, but reinvents what pydantic agents already handle.

Strengths
  • Genuinely thoughtful separation of concerns (Tal/Dance/Act/Stage) reduces mega-prompt brittleness
  • Stage-aware output compilation (MCP vs. GPTs vs. agents) is a clever deployment abstraction
Weaknesses
  • Overlaps with existing agent frameworks (LiteLLM, Anthropic agents) that already decompose control flow
  • Early-stage—website and MCP integration appear minimal; unclear if Stage compilation works in practice
Target Audience

Prompt engineers, LLM application developers building multi-domain systems

Similar To

LiteLLM prompt composition · Anthropic Models Context Protocol (MCP) · Pydantic AI

Similar Projects