$npx -y skills add tobihagemann/turbo --skill review-agentic-setupDetect agentic coding infrastructure in a project: CLAUDE.md, AGENTS.md, installed skills, MCP servers, hooks, and cross-tool compatibility (Claude Code and Codex CLI). Returns structured findings about agentic readiness without applying changes. Use when the user asks to \"revie
| 1 | # Review Agentic Setup |
| 2 | |
| 3 | Detect agentic coding infrastructure and flag gaps across Claude Code and Codex CLI conventions. Analysis only. Does not install or configure anything. |
| 4 | |
| 5 | ## Scope |
| 6 | |
| 7 | Agentic setup review always operates at the project level. Scope parameters (diff commands, file lists) are accepted but ignored. |
| 8 | |
| 9 | When called standalone, use the git repository root as the project root (fall back to the current working directory if not in a git repo). |
| 10 | |
| 11 | ## Step 1: Detect Infrastructure |
| 12 | |
| 13 | Search for agentic coding configuration in the project. Classify findings into five categories: |
| 14 | |
| 15 | ### Agent Instructions |
| 16 | |
| 17 | | File | Read by | What to check | |
| 18 | |---|---|---| |
| 19 | | `CLAUDE.md` (project root) | Claude Code | Exists? Sections present (project structure, conventions, build commands, rules)? | |
| 20 | | `CLAUDE.md` (subdirectories) | Claude Code | Any nested CLAUDE.md files for sub-module guidance? | |
| 21 | | `AGENTS.md` (root and subdirectories) | Codex | Exists? Content areas (setup, conventions, testing, architecture)? | |
| 22 | | `AGENTS.override.md` | Codex | Exists? Overrides AGENTS.md at the same directory level. | |
| 23 | | `README.md` | Both | Mentions AI-assisted development, agent workflows, or prompting guidance? | |
| 24 | |
| 25 | Read each file that exists and summarize its coverage areas. Note which tools can use it. |
| 26 | |
| 27 | Claude Code does not read AGENTS.md directly. Projects share instructions across both tools by: |
| 28 | |
| 29 | - **Import**: CLAUDE.md includes `@AGENTS.md` to pull in shared content |
| 30 | - **Symlink**: `CLAUDE.md` symlinked to `AGENTS.md` (or vice versa) so both tools read the same file |
| 31 | |
| 32 | Check for both patterns when the project has agent instruction files. |
| 33 | |
| 34 | ### Installed Skills |
| 35 | |
| 36 | Check both skill locations. For each skill directory, read the SKILL.md frontmatter to extract `name` and `description`. Group skills by type: |
| 37 | |
| 38 | | Location | Used by | |
| 39 | |---|---| |
| 40 | | `.claude/skills/` | Claude Code | |
| 41 | | `.agents/skills/` | Codex | |
| 42 | |
| 43 | Skill types: |
| 44 | |
| 45 | - **Analysis** — review, audit, or inspection skills |
| 46 | - **Workflow** — multi-step process skills |
| 47 | - **Pipeline** — skills that compose other skills |
| 48 | - **Utility** — formatting, staging, committing, or other focused tools |
| 49 | |
| 50 | Both tools use the same SKILL.md format (YAML frontmatter with `name` and `description`). Projects can symlink one location to the other (e.g., `.agents/skills/` → `.claude/skills/`) to share skills across both tools. Check for symlinks. |
| 51 | |
| 52 | ### MCP Servers |
| 53 | |
| 54 | | File | Used by | What to check | |
| 55 | |---|---|---| |
| 56 | | `.mcp.json` (project root) | Claude Code | Configured MCP servers with their types (stdio, http) | |
| 57 | | `.claude/settings.json` | Claude Code | MCP server entries in project settings | |
| 58 | | `.claude/settings.local.json` | Claude Code | MCP server entries in local settings (gitignored) | |
| 59 | | `.codex/config.toml` (`[mcp_servers]`) | Codex | MCP server entries in project config | |
| 60 | |
| 61 | List each detected server by name, type, and which tool configuration it belongs to. |
| 62 | |
| 63 | ### Hooks |
| 64 | |
| 65 | | File | Used by | What to check | |
| 66 | |---|---|---| |
| 67 | | `.claude/settings.json` | Claude Code | Hook configurations (PreToolUse, PostToolUse, etc.) | |
| 68 | | `.claude/settings.local.json` | Claude Code | Hook configurations in local settings (gitignored) | |
| 69 | | `.codex/hooks.json` | Codex | Hook configurations (PreToolUse, PostToolUse, etc.) | |
| 70 | |
| 71 | List each detected hook with its trigger event, command, and which tool configuration it belongs to. |
| 72 | |
| 73 | ### Documentation References |
| 74 | |
| 75 | Search CLAUDE.md and AGENTS.md content for file paths, URLs, or section headings that reference documentation. Also check the project root for: |
| 76 | |
| 77 | | File or directory | What it indicates | |
| 78 | |---|---| |
| 79 | | `docs/` directory | Project documentation | |
| 80 | | `CONTRIBUTING.md` | Contributing guidelines | |
| 81 | | `TROUBLESHOOTING.md` | Known issues and fixes | |
| 82 | | `ARCHITECTURE.md` | Architecture or design docs | |
| 83 | | API docs or schema files referenced in agent instructions | API reference material | |
| 84 | |
| 85 | ## Step 2: Analyze Gaps and Compatibility |
| 86 | |
| 87 | For each category, assess agentic readiness: |
| 88 | |
| 89 | - **Present and configured** — infrastructure detected with meaningful content |
| 90 | - **Partially configured** — file exists but content is minimal or boilerplate |
| 91 | - **Missing** — no infrastructure detected for this category |
| 92 | |
| 93 | Gap analysis considers the project's complexity. A single-file script doesn't need installed skills. A multi-module application benefits from CLAUDE.md with build commands and architecture notes. |
| 94 | |
| 95 | ### Cross-Tool Compatibility |
| 96 | |
| 97 | After detection, assess which agentic tools the project supports: |
| 98 | |
| 99 | | Signal | Compatibility | |
| 100 | |---|---| |
| 101 | | Has `AGENTS.md` but no `CLAUD |