Visual Inspector MCP has 5 tools with basic schemas and descriptions in Spanish. Tool names follow verb_noun pattern (inspect_html, get_selected_element, highlight_element, apply_css_change, close_inspector). However, descriptions are minimal (avg ~80 chars, below 194-char baseline), lack context on WHEN to use each tool, and omit error recovery guidance. Parameters have types and basic descriptions, but lack constraints (enums, ranges, patterns). No output schemas documented. Error handling is absent, tools return generic text responses without actionable guidance. Security concerns: apply_css_change modifies files without confirmation or dry-run support. Tool composition is reasonable but lacks pagination/batching patterns.

• Document output schemas for all tools. E.g., get_selected_element should return {selector: string, tagName: string, classes: string[], styles: object, html: string}. This enables LLMs to chain tools and extract relevant data.
• Expand descriptions to 100 - 200 chars. Include WHEN to use the tool and what the LLM should do with the result. E.g., 'Highlights an element in the live preview by CSS selector. Use after get_selected_element to visually confirm the right element before applying changes.'
• Add a confirm_css_change tool or dry-run parameter to apply_css_change. Return a preview of the change before writing to disk, allowing the LLM to verify correctness.
• Implement error categorization. Return structured errors: {error: string, category: 'retryable'|'user_fixable'|'fatal', suggestion: string}. E.g., 'Selector not found. Available selectors: [list]. Did you mean one of these?'
• Add constraints to free-form parameters. E.g., 'property' should accept only valid CSS properties (color, font-size, etc.), use an enum or validation with a helpful error message listing valid options.