What problem does it solve? Setting up AI agent configuration (rules, skills, agents, MCP servers) across different tools like Claude Code and OpenCode is error-prone and inconsistent. This Skill automates the integration of agent-core as the single source of truth, generating the correct target artifacts for each platform and verifying the result. ## Core Features & Use Cases - Context and scope detection: Detects whether the project is an existing repo or fresh bootstrap, whether agent-core is installed or a local checkout, and resolves global vs. project scope. - Target-specific generation: Runs sync for Claude Code (CLAUDE.md, .claude/skills, agents, merged settings.json, .mcp.json) and handles the consume-only .tgz bundle workflow for OpenCode (download or build, then extract to ~/.config/opencode/). - MCP wiring and verification: Identifies required environment variable names (e.g., GITHUB_TOKEN) without hardcoding secrets, then verifies each target against a target-specific Definition of Done. - Use Case: A developer clones a new project and wants the team's shared AI agent rules and MCP setup applied. The Skill detects the context, syncs the project-scope Claude Code config, unpacks the OpenCode bundle, and reports which env vars still need to be set. ## Quick Start Ask the assistant to integrate agent-core into the current project at project scope and verify the generated Claude Code and OpenCode configuration.
