ScopeProbe has 3 tools with visible schemas and descriptions, but significant quality gaps limit production readiness. Tool names lack clear action verbs (scope_scan, scope_measure_dc, scope_capture_stats are somewhat descriptive but not verb-first). Descriptions are present but written in Chinese with technical jargon; English descriptions would improve LLM comprehension. Parameter descriptions exist but are minimal (e.g., 'DSCope channel number' for channel). Output schemas are not formally documented, responses are inferred from code (e.g., {ok, voltage, vdiv_mv, ...}). Error handling returns {ok: False, error: str(e)} but lacks recovery guidance or error classification. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite all tools being READ_ONLY. Composition is reasonable, each tool has one responsibility, but no batch variants or pagination support.

• Rename tools to verb-first: 'scan_devices' (or 'list_devices'), 'measure_dc_voltage', 'capture_waveform_stats'. This makes intent immediately clear to LLMs.
• Provide English descriptions (50 - 200 chars each) alongside or instead of Chinese. E.g., 'Scan for connected DSCope oscilloscope devices and return model, serial, and channel info.' Descriptions should state WHAT, WHEN, and WHAT IT RETURNS.
• Document output schemas formally. Add a 'Returns' section to each tool docstring with JSON Schema or a structured example. E.g., scope_measure_dc returns {ok: bool, voltage: float, vdiv_mv: int, code_median: int, code_min: int, code_max: int, clipping: bool, tried: list, hint: str}.
• Enhance error messages with recovery guidance. E.g., 'wavegate-capture not found at $WAVEGATE_CAPTURE. Set the environment variable or check the path in .mcp.json.' or 'No waveform samples captured. Verify the device is connected and the channel is enabled.'
• Add tool annotations: @mcp.tool(readOnlyHint=True) for all three tools to signal safety to LLMs and prevent unnecessary confirmation prompts.