Visible, precise Chrome control for AI assistants, as a Model Context Protocol server. Local by design.
All 6 tools have empty input schemas (no properties, no required fields) and minimal descriptions (10-50 chars). No parameter documentation exists. Tools lack actionable error guidance, output schemas are undocumented, and naming does not follow verb_noun conventions consistently. install_extension, uninstall_extension, and update_apply are write/destructive operations with no confirmation or dry-run support. The server provides no evidence of LLM-optimized descriptions or structured output design.
Get the current status of the Chrome Control MCP browser extension
Install the Chrome Control MCP browser extension
Uninstall the Chrome Control MCP browser extension
Apply available updates to Chrome Control MCP
Check for available updates to Chrome Control MCP
Get the current installation status and version information
All 6 tools have empty input schemas: {"type":"object","properties":{},"required":[],"additionalProperties":false}. Per HARD SCORING RULES, schema score MUST be 0 when no properties are defined.
Descriptions are 10 - 50 characters, below the 10 - 1024 character baseline and well below the 50 - 200 char LLM-optimized range. Per HARD SCORING RULES, descriptions under 20 chars score 0 - 20. Examples: 'Install the Chrome Control MCP browser extension' (50 chars) lacks WHEN to use it, prerequisites, or what it returns.
Destructive operations (uninstall_extension, update_apply) lack confirmation or dry-run support. No error guidance for recovery. Agents cannot distinguish retryable vs fatal errors.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 33 | 2026-07-28+ | v2 |
No output schemas documented. LLMs cannot predict what fields to expect from extension_status, update_status, or update_check. Downstream tool chaining is impossible without trial-and-error.
Naming inconsistency: 'update_status' (read-only) vs 'update_check' vs 'update_apply' creates ambiguity. LLMs may conflate 'update_status' (get info) with 'update_apply' (execute). Recommend: 'get_update_status', 'check_for_updates', 'apply_updates'.