Back to browse
GitHub Repository

Declarative runbook executor for AI agents. Validate, dry-run, and safely run real CLI workflows.

11 starsGo

Define MCP tools as YAML specs

by stevehiehn·Feb 27, 2026·3 points·0 comments

AI Analysis

●●SolidBig BrainSolve My Problem

YAML runbook executor with destructive-step approval gates, exposable as MCP tools directly.

Strengths
  • Destructive-step gating (`--approve` flag) adds real safety guardrail missing from naive agent execution
  • Three step types (shell, HTTP, built-in actions) balance power with structured safety
  • MCP server auto-exports YAML plans as callable tools—no boilerplate wrapper code needed
Weaknesses
  • Narrow audience: only relevant if already using MCP or building LLM agents with tool-use loops
  • Early stage: GitHub shows 3 stars, no visible adoption or production case studies yet
Target Audience

AI/LLM engineers building agentic systems, DevOps automation builders, MCP server developers

Similar To

Anthropic MCP · OpenAI Tools framework · Temporal workflows

Similar Projects

AI/ML●●Solid

Open Agent Spec. Treat AI agents like typed functions not prompt chains

Schema validation catches LLM output mismatches before they break downstream systems.

Big BrainNiche Gem
andrewvector
212mo ago