No. This workflow is not directly supported because favorite modes have explicit account, combination, and incremental-download limitations.

• The README says collect / collectmix work only for the account represented by the logged-in cookies; they are not general favorite-data access modes.
• collect / collectmix must be used alone and cannot be combined with post, like, mix, or music, so they cannot be placed in the same mode configuration as post.
• The README explicitly states that increase applies to post / like / mix / music; favorite collection modes do not support incremental stopping.
• Although number.collect and number.collectmix exist, with 0 meaning unlimited, that does not remove the mode restrictions.

The project can archive favorites for the logged-in account, but it does not directly support a combined-mode, incremental favorite-sync workflow.

No. It cannot transcribe both content types because the README limits transcription to videos; image-note items do not produce transcript files.

• The sample configuration uses model: gpt-4o-mini-transcribe and supports response_formats: ["txt", "json"].
• The Optional Feature section explicitly says the behavior “applies to video items only”; image-note items do not generate transcripts.
• Transcription also requires a valid OPENAI_API_KEY or transcript.api_key; the FAQ says to check transcript.enabled, whether the item is a video, and the selected response formats.
• Transcription runs after video download, so it is a post-processing feature rather than OCR or image-text recognition for image notes.

It is suitable for speech transcription after video archiving. Image notes can still be saved with media and metadata, but the README provides no OCR support.

No. It should not be used directly as a multi-user, highly available backend because the README exposes only a basic job API and local persistence, with no documented authentication, authorization, or distributed task design.

• REST mode starts with --serve --serve-port 8000 and provides submission, single-job status, job listing, and health-check endpoints.
• Completed jobs are pruned by max_jobs and job_ttl_seconds; the defaults are 24 hours and 500 jobs, which is more consistent with a bounded lightweight service.
• The Docker example mounts only config.yml and Downloaded; database_path defaults to a SQLite file in the current working directory.
• Project insights explicitly do not show multi-user support, distributed queues, object storage, permission isolation, or high availability.

It is suitable as a single-host internal service or automation entry point, but a multi-tenant deployment would need additional authentication, isolation, shared storage, and recovery mechanisms.

No. It should not be treated as a stable general-purpose live recorder because the README labels live recording experimental and says HLS normally saves only a playlist.

• Current Limitations states that live recording saves FLV natively, while HLS sources save only the playlist and require ffmpeg for playable output.
• The same section says the webcast room endpoint has not been verified against every live scenario and should be treated as experimental.
• live.* exposes max_duration_seconds, chunk_size, and idle_timeout_seconds, showing that recording controls exist, but not that every stream source is covered.
• Project insights also identify unverified live scenarios and post-processing requirements for FLV/HLS.

It can be used as an experimental collector in controlled cases. For directly playable HLS output, unattended long-running operation, and reliable coverage of all live types, the README does not provide sufficient evidence.

It depends: the project is suitable for batch post archiving, but completeness depends on browser fallback, valid login state, and manual CAPTCHA handling.

• The README explicitly supports post; number.post: 0 means unlimited, while the sample configuration includes thread: 5, retry_times: 3, and SQLite history.
• For receiving only about 20 posts, the README attributes this to pagination risk control and requires browser_fallback.enabled: true, headless: false, manual verification in the browser popup, and keeping the popup open long enough.
• Browser fallback is “fully validated for post”, giving this mode stronger support than like, mix, or music.

This is not an official export tool; the README does not promise complete results for every account, CAPTCHA state, or page change.

Yes. These requirements fit the project’s configuration-driven download, retry, proxy, and disk-based incremental capabilities.

• mode supports post, like, mix, and music; the project description also explicitly lists videos, image notes, and collections.
• The optional proxy field supports HTTP/HTTPS proxies, while retry_times controls retries; the sample configuration sets retry_times: 3.
• With increase.post/like/mix/music: true, existing primary media on disk is skipped; false causes downloads to be redone and overwritten for the current scope.
• The README also provides download integrity checks. Project insights state that checks such as Content-Length are used and incomplete files are removed on failure; SQLite records history, but skipping depends on correctly configuring increase.

It is suitable for recoverable downloads on a single host or a Docker-mounted directory. The README does not guarantee proxy availability, immunity to platform throttling, or a watermark-free source for every item.