agents-best-practices
Active·★ 1.1k·MIT·Updated 2026-05-15
★ Trending★ Multi-Agent★ Dev Tooling
Provider-neutral Agent Skill for Codex, Claude Code, and agentic harness design.
A provider-neutral Agent Skill library for designing, auditing, and refactoring agentic harnesses compatible with Codex and Claude Code. It covers the full control plane of an agent runtime: typed tool design, permission checks, context management, memory, and observability. Targeted at developers building production-ready agent systems across any domain or AI provider.
#agent-skill#agent-skills#agentic-workflows#agents#ai-agents#anthropic#claude#claude-code#multi-agent
01
Features
01Provider-neutral agentic loop design compatible with OpenAI, Anthropic, and compatible APIs
02Typed tool definitions with structured results and runtime permission checks outside the model
03Planning mode and approval-gated execution patterns for safe agent actions
04Context management, memory, and auto-compaction with active state preservation
05Observability, evals, launch gates, and incident response checklists
02
Compatibility
Claude Code
Native
Verified via docs
Codex
Native
Verified via docs
OpenAI API
Supported
Verified via docs
Anthropic API
Native
Verified via docs
03
Quick start
1
$ npx skills add DenisSergeevitch/agents-best-practices -g
04
Use cases
↳Generate MVP agent harness blueprints for any business domain (CRM, ops, finance, healthcare)
↳Audit and refactor brittle or over-permissioned existing agent systems
↳Design narrow typed tools and connector governance for multi-system agents
05
Alternatives
Microsoft AutoGen★ 58.5k
A framework that enables the development of LLM applications using multiple agents that can converse with each other to solve tasks.
CrewAI★ 52.4k
Framework for orchestrating role-playing, autonomous AI agents. By working together, your Crew can tackle complex tasks.
context-mode★ 16.0k
Context window optimization for AI coding agents. Sandboxes tool output, 98% reduction. 12 platforms
Auto-claude-code-research-in-sleep★ 11.0k
ARIS ⚔️ (Auto-Research-In-Sleep) — Claude Code skills for autonomous ML research: cross-model review loops, idea discovery, and experiment automation via Codex MCP
openagent★ 5.1k
⚡️AI Cloud OS: Open-source enterprise-level AI knowledge base and MCP (model-context-protocol)/A2A (agent-to-agent) management platform with admin UI, user management and Single-Sign-On⚡️, supports ChatGPT, Claude, Llama, Ollama, HuggingFace, etc., chat bot demo: https://ai.casibase.com, admin UI demo: https://ai-admin.casibase.com
Related searches
Comments
Log in to leave a comment
No comments yet. Be the first!