Improve MCP discovery and release 0.1.2

[LauraGPT]

Summary

• expand the transcribe_audio tool contract with explicit use/no-use guidance, local path scope, first-call model cache behavior, and output semantics
• bump the MCP server and GHCR image metadata to 0.1.2
• update the container to the current FunASR 1.3.29 runtime
• add regression coverage for the discovery contract and release metadata

Why

Glama's tool discovery quality score evaluates purpose clarity, usage guidance, behavioral transparency, parameter semantics, and contextual completeness. The previous contract named the capability but did not tell an agent when to call it, what inputs are accessible, or why the first call can be slower. Publishing 0.1.2 also triggers a fresh registry/image build for downstream discovery services.

Validation

• python -m unittest discover -s examples/mcp_server -p 'test_*.py' -v (21 passed)
• official MCP Registry JSON schema validation
• mcp-publisher v1.8.0 validate examples/mcp_server/server.json
• Python bytecode compilation for the MCP server and tests
• JSON validation for server.json and glama.json
• git diff --check

The Docker image build and in-container MCP protocol smoke test require Docker, which is unavailable on ind-gpu8; the repository's required Validate and publish FunASR MCP server workflow runs both before merge.

[gemini-code-assist]

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.