babysitter

repository·main·Indexed 23 days ago

https://github.com/a5c-ai/babysitter

An orchestration layer for agentic AI workforces designed to enforce deterministic, hallucination-free workflows. It features the A5C Commander, which utilizes a Kanban-based orchestration model to manage task lifecycles through stages from backlog to production. The system includes tools for defining complex tasks in code, quality gates, human approvals, immutable decision logging, agent stack configuration via the Foundry, and deep execution visibility through a Runs Ledger and Session Forensics.

Tokens
1.4M
Snippets
2.3K
Records
7.3K
Agent score
82%

What's inside babysitter

  1. Overview of DevOps, SRE, and Platform Engineering Specializations

    main

    The DevOps, SRE, and Platform Engineering specialization provides automated processes for cloud infrastructure discovery and cost reduction across AWS, Azure, and GCP.

    Available Processes

    Discovery Processes:

    • aws-systems-discovery.js: AWS account discovery, EKS deep-dive, ALB/NLB attribution, and Route 53/CloudFront exposure analysis.
    • azure-systems-discovery.js: Azure subscription discovery, AKS deep-dive, and AppGW listener-to-backend parsing.
    • gcp-systems-discovery.js: GCP project/org discovery, GKE deep-dive, and GCLB forwarding-rule/backend chain analysis.

    Cost Reduction Processes:

    • aws-systems-cost-reduction.js: AWS cost reduction using Cost Explorer, Compute Optimizer, and Trusted Advisor.
    • azure-systems-cost-reduction.js: Azure cost reduction using Retail Prices API, Azure Advisor, and Cost Management.
    • gcp-systems-cost-reduction.js: GCP cost reduction using Recommender API and Billing BigQuery export.
  2. Overview of API Key Manager

    main
    The API Key Manager is a skill designed for secure API key lifecycle management. It provides a system for generating, rotating, and managing keys with built-in capabilities for usage tracking and quota enforcement. Use this skill when you need to implement API authentication, build management interfaces, or enforce usage limits on specific keys.
  3. Overview of Claude Code Extensibility

    main

    Claude Code is an agentic coding assistant that provides several extensibility surfaces for developers to expand its capabilities:

    • Plugins: Packaged bundles containing skills, agents, hooks, MCP servers, and LSP servers.
    • Skills: Markdown-based instruction sets (SKILL.md) following the AgentSkills.io standard.
    • Hooks: Event-driven handlers (28 event types) that execute shell commands, HTTP requests, LLM prompts, or agentic verifiers.
    • MCP (Model Context Protocol): Servers providing external tools via stdio, SSE, streamable-http, or WebSocket.
    • Subagents: Specialized AI assistants with isolated context, custom system prompts, and independent tool access.
    • Permissions: Fine-grained control over tools using a deny > ask > allow precedence model across managed, user, project, and local scopes.
  4. Overview of Music Album Creation Specialization

    main

    The Music Album Creation Specialization is designed for AI-assisted music generation on text-to-music platforms like Suno and Udio. It provides structured workflows to manage the full creative pipeline, including artist persona development, album conceptualization, song composition, musical style specification, and visual art direction.

    This specialization is intended to produce production-ready content, including lyrics with embedded production notes, detailed style descriptions for AI prompts, and text-to-image prompts for artwork.