Settings

Use daily for the normal devotional path or custom_story for operator-seeded runs.

Run-level prompt bundle identifier carried through resume and publish.

Primary visual composition engine.

Composition id passed to the background worker as remotion_composition_id. Must match a Composition registered in the worker Remotion project.

glass_header: centered frosted-glass title card. compact_header: top-left stack (brand, title, scripture, gold line) — closer to minimal devotional plates.

Subtle animated gold specks over the background. Turn off for the cleanest look on bright footage.

Default publish tagline shared into downstream packaging.

Keep false unless you want experimental shadow-planning behavior.

How many symbolic blueprint candidates the planner scores before selection.

Target weighting for fresh episodes versus continuations.

Target weighting for continuation episodes when real unresolved threads exist.

Hard duplicate protection window for exact symbolic fingerprints.

How far back novelty coverage counts are considered.

Reject near-duplicate symbolic blueprints at or above this overlap score.

How similar a new voice/script package can be before the system treats it as stale.

Minimum spoken realism score expected before voice/render proceeds.

Minimum channel-fit score for a selected story space.

How many times WF01 may rework script output after guardrail failures.

How many times WF02 may revise spoken performance after voice gates fail.

Model id sent to ElevenLabs cast synthesis. Flash v2.5 is the fastest, Eleven v3 is the most expressive, and Turbo v2.5 remains the balanced default.

Enable the Upload-Post lane in WF03. When upload_post_platforms is blank, legacy behavior claims all enabled publish targets through Upload-Post. When upload_post_platforms is set, Upload-Post only handles that explicit list and direct branches can run in parallel.

Upload-Post account/profile handle used by the WF03 Upload-Post node when that branch is enabled.

Comma-separated platform list for the Upload-Post lane, for example youtube,instagram,tiktok. Leave blank to preserve legacy XOR behavior. Set this explicitly when you want Upload-Post-only, direct-only, or both for a platform such as YouTube.

Enable the direct YouTube branch in WF03. Combine with upload_post_platforms when you want both direct YouTube and Upload-Post to run.

Enable the direct Instagram branch. Leave false when Upload-Post alone should handle Instagram.

Enable TikTok as a publish target. The direct publish queue still blocks TikTok, so Upload-Post is the expected production path here.

Enable the direct Facebook branch in WF03.

Enable X as a publish target. The direct publish queue still blocks X, so Upload-Post is the expected production path here.

Used by the direct Instagram branch. Leave blank when Upload-Post is handling Instagram instead.

Used by the direct Facebook branch and as a page-selection hint in publishing payloads.

Auto switches between comedy and everyday-faith rotation tags. Force a mode only when you want to override the content-driven choice.

Core reach tags added to every post. Comma or newline separated.

Used for lighter, banter-driven, or skit-leaning posts. Comma or newline separated.

Used for grounded reflection, testimony, or straight devotional posts. Comma or newline separated.

Optional manual add-ons layered after the core and rotation tags. Comma or newline separated.

Primary video aspect ratio passed into render composition.

Spoken program target in seconds (aim for this length excluding outro). WF01 Scriptwriter and WF02 Voice Writer should budget dialogue to this window so the short cut is not doing emergency surgery on meaning. Not a hard ffmpeg cap: final render follows finished audio.

Which turn list the media worker uses: full dialogue vs short-cut abridged lines. Use full while tuning scripts so lines are not dropped mid-conversation; auto may pick short for vertical shorts when estimates pass gates.

Enable or disable the opening title card overlay.

Enable or disable spoken subtitles in the output render.

Worker-side quality hint for final render output.

How long the opening title card remains before spoken captions take over.

Fade-in duration for the opening title treatment.

How the worker selects and uses background footage.

Whether local assets or remote assets should be preferred first.

Optional absolute path to force a specific background loop asset.

Primary caption style mode for the render composition.

Special handling for hook words and opening captions.

How scripture or closing landing phrases are visually emphasized.

Hard ceiling for simultaneous visible words in a caption burst.

Upper bound for each subtitle line before wrapping.

Maximum number of subtitle lines rendered at once.

Safe margin ratio used for portrait subtitle placement.

Primary subtitle vertical placement zone.

Allow emphasized words and active-word highlight behavior.

Overall motion intensity for caption entrances and emphasis.

Hint used by downstream creative steering to balance density.

Optional absolute path to the default outro image.

How long the standard outro should stay on screen in seconds.

Fade duration for the standard outro transition.

Use inherit to mirror the standard outro, or force on/off for short cuts.

Optional override image for short-cut outro renders.

How long the short-cut outro should stay on screen in seconds.

Fade duration for the short-cut outro transition.

Tail fade duration for background music when the program audio ends.

Base dialogue-under music gain sent to the worker before ducking.

Lower the music bed automatically while dialogue is active.

Sidechain threshold that decides when dialogue should push the music bed down.

Compression ratio used while dialogue is present.

How quickly the music ducks when speech starts.

How slowly the music recovers after speech stops.

Makeup gain applied after ducking compression.

Optional absolute path to force a specific music bed.

Recurring music bed: absolute path, or path relative to storage_root on the worker (e.g. music/track.mp3). When ambient_volume is 0 or no primary ambient file is staged, the media worker mixes this bed under dialogue so Remotion is not voice-only.

Restrict selection to the signature track when true.

Absolute base path on the VPS for resolving relative signature_music_local_path (must be forwarded on the render-podcast JSON body for the worker).

Overrides for shaping custom-story seeds into world-pulse fuel.

Overrides for editorial routing and emotional-center decisions.

Overrides for cast continuity, discourse mode, and selected character burden.

Overrides for blueprint expansion and story-brief shaping.

Overrides for hook framing, delivery logic, and entry posture.

Overrides for dialogue scaffold and captions. Prefer rules that force reactive back-and-forth (answer the last line, interrupt, callback), grounded sensory detail over cartoon exaggeration, and spoken length that fits target_duration.

Overrides for scripture and context research framing.

Overrides for theological review of the research packet.

Overrides for theological and craft review of the final script package.

Overrides for spoken-script lexicalization, line feel, and final episode speech shape.

Overrides for the unified voice-evaluator and spoken-realism review pass in WF02.