This server has a single tool 'execute_skill_action' with critically inadequate specification. The tool description in mcp_server.py is minimal ('Execute a skill action for the dual-track vector quantized acoustic codec'), 79 characters, meeting the length floor but providing no actionable guidance on when/why to call it, what parameters it accepts, or what it returns. No input schema is visible in the source code. The mcp_server.py file only prints a JSON initialization message with tool names; there is no actual tool registration with JSON Schema, parameter definitions, or output schema documentation. The client.py file shows a private implementation detail (a DualTrackVectorQuantizedAcousticCodecClient class) but does not expose tool parameter contracts or schemas. The rubric's hard scoring rule states: 'If a tool has NO input schema at all: its schema score MUST be 0.' and 'If you cannot see the actual tool definition in the source (only inferred): cap that tool's overall at 50.' Here, we infer the tool exists from the printed JSON, but we cannot see any explicit schema registration in the source.

• Define the tool with a proper JSON Schema input and output schema. Specify all parameters (audio_duration_seconds, sample_rate_hz, codebook_quantization_levels) with types, descriptions, and constraints (e.g., audio_duration_seconds: 0.1 - 3600 seconds, sample_rate_hz: enum [8000, 16000, 22050, 44100, 48000], codebook_quantization_levels: 1 - 16).
• Expand the tool description to 100 - 200 characters explaining the purpose: 'Encode and quantize audio using a dual-track vector quantized acoustic codec. Returns codec job ID, bitrate, PESQ fidelity score, and timbre preservation metrics.'
• Add a description to every parameter. For example: 'audio_duration_seconds (float, 0.1 - 3600): Duration of audio to encode in seconds. Default: 10.0.'
• Document the output schema. List all return fields with types: codec_job_id (string), sample_rate_hz (integer), bitrate_kbps (float), quantized_tokens_count (integer), pesq_acoustic_fidelity_score (float, 0 - 5), zero_shot_timbre_preservation_pct (float, 0 - 100).
• Rename the tool to a more specific action verb. Consider 'encode_audio_codec' or 'quantize_audio_latent' to clarify the operation.
• Add error handling and recovery guidance. Document what errors can occur (invalid duration, unsupported sample rate, codec failure) and how the LLM should respond (retry with different parameters, call a diagnostic tool, etc.).
• Implement proper MCP tool registration with InitializeResult or tool definition messages, not just a JSON print statement. Use the official MCP protocol for tool announcement.