AutoVideo-Agent v0.2.0

Highlights

• Provider-based media pipeline
• Experimental ComfyUI provider
• Mock / Command TTS
• Scene-level SRT
• Timeline validation
• FFmpeg mixed-media renderer
• QA PASS/WARNING/FAIL
• PyPI distribution
• Trusted Publishing

Validation status

The ComfyUI Provider remains Experimental. It currently has mocked integration coverage only; live workflow validation is still tracked in Issue #12.

AutoVideo-Agent v0.2.0-beta.1

Highlights

• Provider architecture with normalized MediaArtifact and AudioArtifact contracts
• Experimental ComfyUI API provider with prompt submission, history polling, retry, timeout, resume, node mapping, and output download
• Mock and Command TTS providers
• Scene-level SRT subtitles based on actual audio duration
• Continuous timeline validation
• Mixed image/video FFmpeg renderer with explicit trim/loop behavior
• Deterministic QA with PASS/WARNING/FAIL
• v0.1 offline compatibility preserved

Testing

• 34 pytest tests passed (31 baseline tests plus review regression coverage)
• GitHub Actions PASS on main
• Offline regression PASS
• ComfyUI API mocked integration PASS

Known limitation

No live ComfyUI workflow has yet been validated for this prerelease. The ComfyUI provider is implemented but Experimental and awaiting live workflow validation; it is not production-ready.

Call for testing

Real ComfyUI users are invited to validate an image or video API-format workflow and report ComfyUI version, workflow type, and reproducible credential-free configuration in Issue #12. Do not submit API keys, private server credentials, usernames, passwords, private workflows, or model paths.

AutoVideo-Agent v0.1.0

First public release of the local-first Markdown-to-video pipeline. Includes the autovideo CLI, storyboard parsing, deterministic scene assets, silent audio timeline, FFmpeg MP4 rendering with graceful degradation, tests, examples, bilingual documentation, and GitHub Actions CI.