v0.1.0 — script to voiceover + captions

Added

• stdio MCP server miragea-voice-mcp, runnable with uvx --from git+https://github.com/RTY2345RR/miragea-voice-mcp miragea-voice-mcp (PyPI release uvx miragea-voice-mcp coming soon).
• Tools: miragea_get_capabilities, miragea_list_voices, miragea_text_to_speech,
  miragea_transcribe, miragea_script_to_voiceover, miragea_get_task, miragea_list_tasks.
• One-shot script → voiceover + SRT/VTT captions, with script-accurate caption wording and
  short-form cue splitting (max_caption_chars).
• Local WebVTT rendering from Miragea's timed segments.
• No duplicate jobs on retry: idempotency keys on every job, automatic retries reuse the same key,
  timeouts return a resumable task_id, pre-flight validation against live capabilities.
• --check (read-only key/API check) and --print-config for Claude Desktop, Claude Code, Cursor,
  Codex, VS Code and Windsurf.
• Compatible with MCP Python SDK 1.x (FastMCP) and 2.x (MCPServer).
• Companion Agent Skill skills/miragea-voiceover.
• Registry metadata: server.json, smithery.yaml, glama.json, llms.txt.

Install (PyPI release coming soon):

uvx --from git+https://github.com/RTY2345RR/miragea-voice-mcp@v0.1.0 miragea-voice-mcp --print-config claude-code

Live demo: https://huggingface.co/spaces/miragea/miragea-voice-mcp