VS Code extension for visual Arduino & MicroPython programming with Blockly. Multi-board support, project-local Agent Skills, AI camera integration, PlatformIO integration, and 15 languages.
Scoring was not performed
Tool descriptions are minimalist (10 - 40 characters) and lack context. They state WHAT the tool does but omit WHEN to use it, prerequisites, and how return values flow to downstream tools. The 72-char average baseline for production tools is not met. Example: 'Query block definition by type name' (33 chars) vs. 'Query the Blockly block definition for a specific block type (e.g., "motor_forward"). Returns the block's input parameters, output type, color, and tooltip. Use this before validate_workspace to check if a block is available on the target platform.' (212 chars with context).
Output schemas are not documented in the code. The input schemas show parameter types and descriptions, but there is no visible documentation of what fields each tool returns, their types, nested structure, or how to chain tool outputs into downstream tool inputs. Production baseline is 100% of A+ tools have documented return types.
Parameter descriptions are missing or extremely brief. The input schema for 'blockType' says 'The block type identifier' (28 chars) but does not explain format, valid values, enumeration, or where to find valid block type names. Same issue across all tools. Production baseline is 100% of A+ tools have descriptions on ALL parameters.
No enumeration constraints visible for parameters that should be constrained. 'boardName' in get_platform_config accepts 'uno', 'esp32', 'cyberbrick' but these are described as examples in package.json, not as enum schema constraints in the tool definition. Free-form strings invite hallucinated values.
No error handling guidance visible. If query_block_by_type is called with an invalid blockType, what error is returned? What should the LLM do next? Should it call search_blocks to discover valid block types? No recovery path is documented.
No tool composition guidance. The tools are independent (query, search, list discovery vs validate, generate for workspace operations) but there is no documentation of the intended workflow. Should an agent call list_block_categories first? Should it call get_platform_config before validate_workspace? The discovery and execution phases are not connected.
generate_code has two parameters (workspaceXml, language) but no documentation of the expected XML structure, valid language values, code output format, or error cases (e.g., invalid XML, unsupported language).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 0 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 39 | - | v1 |