Back to Ecosystem Pulse
ToolProduction ReadyMCP
agents
by wshobson
Marketplace of portable agent skills and plugins for multiple agent CLIs
Python
Updated Jun 29, 2026
Share:
Overview
Provides a multi-harness plugin marketplace for agent CLIs like Claude Code, Codex CLI, Cursor, OpenCode, and Gemini CLI. Organizes agent skills and plugins so developers can swap runtimes and test the same orchestration across different LLM backends. Distinctive for its cross-LLM plugin adapter approach and focus on reusable "skills" that run against multiple agent CLIs. This ecosystem aligns with the Orchestrator-Worker Pattern.
Key Benefits
As teams build multi-agent stacks, being able to exercise the same skill set across different LLM runtimes reveals brittle behaviors and divergent outputs. Until now, swapping LLM backends or CLIs meant rewriting glue code; a marketplace of portable skills makes comparative testing and iterative hardening practical. That visibility is valuable for multi-agent trust and agent-to-agent evaluation because it surfaces where agents disagree, fail, or delegate poorly. This perspective connects to the Model Context Protocol (MCP) Pattern and the Mutual Verification Pattern.
Ideal For
Developers and researchers who need to prototype, compare, and reuse agent skills across multiple LLM CLIs and runtimes. This approach complements established templates like the Role-Based Agent Pattern.
Real-World Examples
- Evaluate the same agent skill across different LLM backends to spot behavior drift
- Prototype agent workflows by composing reusable plugins that run on multiple CLIs
- Compare agent delegation and failure modes between Claude, Codex, Gemini, and Cursor
Works With
anthropicopenaigooglecursor
Topics
agent-skillsagentic-aiagentsai-agentsanthropicautomationclaude-codeclaude-code-pluginsclaude-code-skillscodex-cli+10 more
Similar Tools
autogencrewai
Keywords
multi-agent orchestrationagent-skillsmulti-agent trustanthropic