Changelog

Track changes to the PowerPost API. Subscribe to updates via webhooks or check this page periodically.

---

September 2026

• Instagram AI label — Publish and schedule accept is_ai_generated on instagram-feed, instagram-reel, and instagram-story. It defaults to off. Set it to true to add Instagram's AI info label. Studio has the same switch under Publishing Settings. See Publish settings.

• Spoken pill shadow — A spoken chip with no highlight.style.shadow no longer draws the cue shadow on that word. The cue shadow still shows outside the window. See Video captions.

• Spoken word captions — A cue highlight sets the spoken color, pill, and in-place scale. A span highlight sets when that word uses it. style still lasts the whole cue. Same render endpoint. See Video captions.

• List voices — GET /api/v1/audio/voices returns the speech voices you can pass as voice_id. Scope is audio:read. No workspace header. MCP tool: powerpost_list_audio_voices. See Audio generation.

• Audio requests omit the provider — Do not send provider on POST /api/v1/audio/generate. Generation status no longer returns it. Older requests that still send the previous value keep working.

• Sound effects and music — Audio generation now accepts mode: "sound_effect" for effects and ambience or mode: "music" for music. Both require duration_seconds; speech remains the default. Uses the existing generation/status endpoints, storage, and refunds. Available in Studio and MCP. If a music prompt looks copyrighted, the failed job includes a suggested rewrite when one is available. See Audio generation.

• RTL video captions — style.text.direction is ltr (default) or rtl. Set rtl for Arabic or Hebrew. Burns then use libass Encoding: -1 so the whole line, including colored spans, is shaped as one RTL run. Spans inherit the cue and cannot set direction. See Video captions.

• Audio Generation API — Generate speech through POST /api/v1/audio/generate, then poll GET /api/v1/audio/generations/{id}. Generic fields stay at the top level and mode-specific settings live under provider_options. Completed files are stored in the workspace media library as media_type: audio (audio.media_id); they cannot be attached to posts yet. Needs a prior credit purchase (MEDIA_REQUIRES_PURCHASE), same as image and video generation. See Audio generation.

---

August 2026

• Caption highlight chips include stroke — A span background grows by stroke.width so the outline stays inside the chip. See Video captions.
• MCP captions match the REST API — powerpost_generate_content and powerpost_render_video_captions advertise the same request fields as the public endpoints, including video-caption style, spans, scale animations, and font size 10–320. See MCP and Video captions.
• Image and video generation require a purchase — POST /api/v1/images/generate and POST /api/v1/videos/generate return 402 MEDIA_REQUIRES_PURCHASE until the account has bought credits. Signup credits still work for writing and publishing. See Errors.
• Threads — threads-post is a first-class platform: generate, publish (text, image, video, carousel), comments, and post analytics. One topic tag comes from the first #tag in the caption. Optional reply_control on publish/schedule (everyone by default). See Post types and Publish settings.

---

July 2026

• Post type limits (docs + validation) — tiktok-photos accepts 1–35 images (was documented as 2–35). Caption length checks match platform rules: X uses weighted length (emoji/CJK = 2, URLs = 23); YouTube descriptions use UTF-8 bytes (5,000). Instagram/Facebook Stories are media-only, so content is optional there and ignored if sent. See Media, caption, and title limits.
• Regenerate Content — POST /api/v1/content/generations/{id}/regenerate regenerates caption content for one platform on a completed generation, with optional free-text refinement feedback (e.g. "be more friendly"). Uses credits. MCP tool: powerpost_regenerate_content. See Regenerate content.
• List Published Posts — GET /api/v1/post-items returns published post items filtered by platform (one or many), a date-only range on posted_at (YYYY-MM-DD), and limit. MCP tool: powerpost_list_post_items. See List published posts.

---

May 2026

• Publish Settings on Publish & Schedule — POST /api/v1/posts/{id}/publish and POST /api/v1/posts/{id}/schedule now accept an optional items[] array with per-item settings (TikTok privacy, YouTube visibility). Items omitted from the array publish with defaults (TikTok SELF_ONLY, YouTube private). See Publish settings reference.

---

April 2026

• Post Item Analytics — GET /api/v1/post-items/{id}/analytics returns the latest views, likes, and comments snapshot for a post item. Each item is polled once a day for up to 90 days after posting; check fetched_at to see how fresh the numbers are.

---

March 2026

• Video Generation API — Generate AI videos via POST /videos/generate with text-to-video and image-to-video support. Three models available: Kling 3.0 Pro, Veo 3.1, and Runway Gen-4.5
• Source URLs — Pass source_urls in content generation requests to give the AI website context for better captions
• Video Webhooks — video_generation.completed and video_generation.failed events now available

---

February 2026

v1 — Initial Release

The PowerPost API v1 is now available. This release includes:

• Content Generation — Generate platform-optimized captions from text, images, or video
• Image Generation — Create AI images from prompts, references, or existing captions
• Media Upload — Upload images (JPEG, PNG, WebP) and videos (MP4, MOV)
• Posts & Publishing — Create posts and publish to Instagram, TikTok, X, YouTube, Facebook, and LinkedIn
• Credit System — Pay-per-use billing with per-operation credit costs
• Webhooks — Receive notifications when generation and publishing jobs complete
• Deep Research Mode — Extended research for higher engagement content
• Custom Writing Styles — Apply your brand voice via the API