Content
31%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a sprawling monolith: concrete CLI commands are present, but they are buried in heavy padding (a 'What is MCP?' primer, dozens of repetitive YOLO prompt examples) and mixed with non-executable code (invalid JSON comments, missing imports, calls to undefined methods). Batch/destructive workflows lack validation checkpoints, and no reference files exist despite abundant content that clearly belongs in them.
Suggestions
Cut the 'What is MCP?' primer and collapse the repetitive `gemini --yolo -p` example blocks into one pattern reference; keep the body to the core workflow (list/add/remove/configure) and move full server implementations, security, and performance content into references/ files.
Fix the code examples so they are executable: remove `//` comments from JSON, import `os` and return results in the Python server, and correct the Node example's import/class usage.
Add explicit validation checkpoints before batch or destructive steps, e.g. run `gemini mcp test <server>` after add/update and verify status before and after 'restart-all'.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | A ~780-line body opens with a 'What is MCP?' section explaining a protocol Claude already knows, then pads every section with repeated `gemini --yolo -p "..."` prompt variations and near-duplicate secure-server JSON configs. This is squarely anchor 1: 'Severely verbose; extensively explains concepts Claude already knows; heavily padded'. | 1 / 5 |
Actionability | Concrete CLI commands exist (`gemini mcp list/add/remove`, config file paths), but many examples would not run: JSON blocks contain `//` comments (invalid JSON), the Python server references `os.environ` without importing `os`, never returns from `handle_request`, and calls `server.start()` on a class defining no `start`; the Node example imports a nonexistent `@modelcontextprotocol/server` `MCPServer` class. This matches anchor 3 ('pseudocode instead of executable code; missing key details') rather than 4, whose examples are executable with only minor gaps. | 3 / 5 |
Workflow Clarity | The Quick Start (list -> add -> remove) is clearly sequenced and troubleshooting steps exist, but batch and destructive operations ('Safely restart all MCP servers in dependency order', `restart-all`, server removal, `--yolo` write workflows) lack any validate-then-proceed checkpoints. Per the guideline capping batch/destructive workflows without validation at 3, this fits anchor 3 ('sequence present but checkpoints missing or implicit') and cannot score higher. | 3 / 5 |
Progressive Disclosure | No references/, scripts/, or assets/ exist — everything (full server implementations, security, audit logging, load balancing, troubleshooting, performance tuning) is inlined in one monolithic 780-line file. That is anchor 2 ('content that clearly belongs in separate files is inlined'); the section headers alone do not reach anchor 3, which still expects meaningful external reference structure. | 2 / 5 |
Total | 9 / 20 Passed |