Agent
Default orchestrator agent. Use for: planning features, reviewing architecture, managing branches/PRs, coordinating work across agents, documentation, debugging non-code issues, all tasks not specifically delegated to another agent.
Agent
Use when: implementing mcp-homelab tool functions, wiring SSH/API logic, writing Python code for the MCP server, debugging tool implementations, adding new tools, refactoring core modules, writing tests.
Agent
Use when: updating README, guides, deployment docs, design docs, backlog files, archiving completed docs, cleaning stale references, writing operational runbooks, updating CONTRIBUTING.md, auditing doc accuracy against current code state.
Agent
Use when: reviewing Codex-produced code before merge, auditing modules for design pattern compliance, evaluating scalability and extensibility of new tool implementations, checking type correctness and layer separation, reviewing any Python change for engineering quality (not security, not docs).
Agent
Use when: reviewing security-sensitive changes (auth, networking, SSH, API endpoints, secrets management), evaluating infrastructure decisions, auditing code for vulnerabilities, or before committing any change that touches authentication, transport, or access control.
Instructions file GitHub Copilot
Instructions for asvarnon/mcp-homelab, covering agent workflow instructions, session start (do this before anything else), session notes (during the session), promote findings immediately and release / publish checklist.
Instructions file GitHub Copilot
Instructions for asvarnon/mcp-homelab, covering mcp-homelab code review instructions, project summary, tech stack, design principles (enforce these in review) and layer separation.
MCP server Claude CodeCodexCursor
MCP server that exposes homelab infrastructure tools to AI assistants. Runs locally from the mcp-homelab Python package.