AgentsMesh is an open-source agent harness and orchestration framework designed to standardize, validate, and execute portable agent skills across heterogeneous AI coding agent environments. In day-to-day operations, AgentsMesh maintains a canonical repository of skill definitions (`agents/skills/<name>`) and projects them via symlinks into client-specific directories such as `.claude/skills` and `.agents/skills`. This design allows CLI coding agents like Claude Code and OpenAI Codex to discover and execute identical automation workflows. The framework integrates with the Model Context Protocol (MCP) to hook agents into external tools, such as Chrome DevTools for automated browser verification and specialized Git workflow automations. What distinguishes AgentsMesh is its contract-driven validation harness. Rather than locking repository workflows to a single agent vendor, it includes automated validation scripts (`agents/validate.sh`) and CI pipeline checks (such as Bazel-based contract gates) to enforce frontmatter portability, path integrity, and client-neutral tool definitions across all agent capabilities. AgentsMesh is free and open-source software distributed via GitHub, maintained with automated container builds and CI integration for scalable agent orchestration.
Tags: agent harness, ai agent, cli, coding agent, developer tools, OpenSource