Content
57%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.
A well-structured, largely actionable reference for calling Gemini CLI, with solid executable command and config examples. Its main weaknesses are token padding from duplicated/narrative prose, no validation feedback loops around the risky YOLO mode, and no progressive disclosure into separate reference files.
Suggestions
Remove the 'When to Use' section and first-person workflow narrative lines ('I'll use Gemini...', 'Let me get...') since they duplicate the frontmatter description and add tokens without new guidance.
Add an explicit validation/feedback step around YOLO-mode invocations (e.g. review Gemini's proposed file changes before applying, or run in sandbox first) to satisfy the destructive-operation workflow cap.
Move the bulkier material — full MCP setup config, the example-workflow catalog, and the three authentication methods — into references/ files (e.g. AUTH.md, EXAMPLES.md) linked one level deep from a leaner SKILL.md overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient (tables, executable commands, code blocks) but includes unnecessary padding such as the 'When to Use' list that duplicates the frontmatter description and first-person narrative prose in example workflows ('I'll use Gemini for real-time search:', 'Let me get Gemini's perspective on this code:'), so it fits 'mostly efficient but could be tightened' rather than the lean anchor 4. | 3 / 5 |
Actionability | Provides largely copy-paste-ready guidance: concrete `gemini -p` CLI invocations, a real MCP tool list, a Rust snippet, and a JSON config block, matching 'mostly executable; minor gaps'; not a 5 because the example workflows use pseudo bracket notation like `[gemini_search]` / `[gemini_prompt_with_input]` that is not literal executable syntax. | 4 / 5 |
Workflow Clarity | Content is organized by usage method (MCP / CLI / Rust) with supporting config, error handling, and auth sections, giving a rough sequence, but there are no explicit validation checkpoints or feedback loops — notably for the risky 'YOLO mode (auto-approve all actions)' path, which per the rubric cap for destructive/auto-approved operations holds this at 3. | 3 / 5 |
Progressive Disclosure | The body has clear section headers but no bundle files exist (references/, scripts/, assets/ absent) and all content — full config, multiple example workflows, and three auth methods — is inlined in one ~160-line file, fitting 'some structure but content that should be separate is inline'; the under-50-line simple-skill exception does not apply at this length. | 3 / 5 |
Total | 13 / 20 Passed |