兼容任何MCP客户端ClaudeCodexChatGPTCursorGeminiOpenClaw
托管MCP服务器

一个端点。 完整视频编辑器。

让任何支持MCP的代理直接访问您的BlitzReels工作区。它可以检查媒体、编辑时间线、运行工作流、生成素材并导出成品视频。

blitzreels / mcp直播
127
规范工具
OAuth
工作区认证
HTTP
可流式传输
端点
https://blitzreels.com/api/mcp
示例MCP配置
{
  "mcpServers": {
    "blitzreels": {
      "type": "http",
      "url": "https://blitzreels.com/api/mcp"
    }
  }
}

首次调用会打开 OAuth 登录。代理以您的身份操作,权限限定在您的工作区。

三步完成

从您已使用的客户端连接。

端点和功能在Claude、Codex、ChatGPT、Cursor及其他MCP客户端间保持一致。

  1. 01

    添加端点

    将托管的BlitzReels URL粘贴到客户端的MCP服务器或连接器设置中。

  2. 02

    批准OAuth

    一次登录。所有工具调用均限定在您的BlitzReels工作区和权限内。

  3. 03

    描述编辑内容

    让代理检查、编辑、生成或导出。它从相同的规范工具目录中选择。

实时功能索引

全部127个MCP工具。

名称和描述来自提供给连接代理的相同描述符,确保列表与发现保持一致。

阅读文档
127 工具

项目和导出

27

List Projects

projects_list
读取

List the user's BlitzReels projects with status, duration, and basic metadata.

Get Project

projects_get
读取

Get details for one BlitzReels project, including timeline summary, clips, captions, and current recovery state with an app path and next read tool.

Inspect Project

projects_inspect
读取

Read bounded project context for editing, including timeline items, media assets, transcripts, captions, and stable IDs.

List Exports

exports_list
读取

List exports with render status and progress. Completed exports include a signed download URL. Pass exportId to return one export.

Read Export Caption Track

exports_caption_track_get
读取

Read a WebVTT caption track from the completed export's immutable canonical speech snapshot, preserving the actual export trims. Returns unavailable for older exports or unverified speech; never fabricates captions from titles or current edited timelines.

editor asset replace

editor_asset_replace
写下

Replace one shot's media asset while preserving the other timeline items. Revision checked; dryRun previews by default. Requires a current editor snapshot and an asset in the same workspace.

editor audio set

editor_audio_set
写下

Adjust one item's volume and audio fades. Revision checked; dryRun previews by default. Preserves the rest of the project.

editor text set

editor_text_set
写下

Create or revise one title or text overlay with explicit timing. Revision checked; dryRun previews by default. This edits project text, not spoken audio or captions.

editor crop set

editor_crop_set
写下

Adjust the crop of one existing source view after a layout has been applied. Revision checked; dryRun previews by default. Preserves the other shots.

editor assets insert

editor_assets_insert
写下

Insert a workspace image, generated shot or audio asset at an explicit project time. Revision checked; dryRun previews by default.

editor items move

editor_items_move
写下

Move selected timeline items to explicit times and layers. Revision checked; dryRun previews by default.

editor items remove

editor_items_remove
写下

Remove selected timeline items and, if requested, their linked captions. Revision checked; dryRun previews by default. The saved action can be undone.

Read Editor Snapshot

editor_snapshot_get
读取

Read the current revision, stable timeline IDs, captions, style and bounded media context before a precise revision. Signed media links are refreshed by reading again.

List Edit History

editor_history_list
读取

Read up to 50 saved edit actions, their labels and undo state. Does not restore or change the project.

Undo Saved Edit

editor_edit_undo
写下

Restore one saved editor action at the expected current revision. Requires the action ID from edit history and an idempotency key. Existing project data is changed.

Review Clip Candidates

clips_candidates_list
读取

Read existing suggested moments with source ranges, hooks, transcript excerpts and thumbnails before producing videos. Results are bounded; this does not generate suggestions or render clips.

Read Caption Words

captions_words_list
读取

Read bounded editable caption words and stable word IDs for one project, optionally matching text.

Correct Caption Words

captions_words_correct
写下

Replace selected caption words by their stable IDs. Preserves timing and the other captions; this does not change the spoken audio.

Save Video Quality Feedback

video_feedback_save
写下

Save the user's 1-5 rating and optional quality category/comment against one completed playable video export. Replaces that user's previous feedback for this export; does not change accepted clips or defaults. Requires the feedback storage migration to be installed.

Prepare Source Captions

captions_source_prepare
写下

Restore editable project captions from the selected source's completed canonical transcript. Keeps the full recording and avoids duplicate captions on identical retries. Requires source transcription first; no-speech is an explicit successful state.

Recover Missing Clips

clips_recover
写下

Recover a terminal batch's missing clip count through the persisted recovery workflow. Preserves accepted clips and returns the recovery batch ID. Can spend existing account credits; retry with the identical idempotency key.

List Workspaces

workspaces_list
读取

List bounded workspaces the connected user belongs to, with their role and default workspace. Selecting a workspace for the task does not change account defaults.

Read Business Context

workspace_context_get
读取

Read the selected workspace's business audience, offer, reviewed brand defaults, protected vocabulary, editorial purpose, language, avoided terms, pronunciation notes and existing account entitlements. Returns current export quality and watermark constraints without pricing or upgrade links.

Save Reviewed Business Context

workspace_context_save
写下

Save the user-reviewed business name, audience, offer, conversion goal, tone, protected vocabulary and editorial preferences for the selected workspace. Requires workspace settings permission and confirmReviewed=true; changes future assistant guidance, preserving existing projects. Pronunciation notes guide script drafting; they do not retime or replace source speech.

Inspect Supplied Business Website

workspace_brand_scan
写下

Start extracting a brand profile from the exact website the user supplied and authorized to scan. Saves a draft profile for review; does not apply discovered fields to existing videos. Can spend existing account credits.

Read Website Brand Draft

workspace_brand_scan_get
读取

Read one authorized website extraction run and its draft business profile. Show the extracted fields to the user for review before saving them as defaults.

Read Published Content Performance

content_performance_list
读取

Read bounded existing provider-reported engagement in the selected workspace, sorted by views. Returns source title, caption, published date and views/likes/comments/shares/saves. Does not import new data, start monitoring or infer attributable leads, sales or revenue.

媒体库和上传

15

List Media

media_list
读取

List media files in user's library (videos, audio, images)

Get Media Transcript

media_transcript_get
读取

Get transcript summary metadata or a bounded window of transcription segments for a video asset.

Upload File

media_upload_file
写下

Import a video, audio, or image file from ChatGPT into the BlitzReels media library. Use this for a file the user attached or picked, and for an image ChatGPT generated in this conversation that the user wants to keep, animate, or edit in BlitzReels.

Import Media From URL

media_import_url
写下

Import a video, audio, or image from a user-provided direct URL into the private BlitzReels media library.

Start Media Upload

media_upload_start
写下

Get a presigned URL for direct file upload (for large files)

Finish Media Upload

media_upload_finish
写下

Complete the upload process after uploading to a presigned URL

Inspect Import URL

media_import_inspect
读取

Inspect a user-provided public social-video or Google Drive URL and return available import metadata without creating media.

Scan Page For Video

media_import_scan_page
读取

Scan a user-controlled public page and return an importable video URL found on it without creating media.

Get Media Asset

media_assets_get
读取

Get metadata and processing state for one media-library asset without returning transcript segments.

Preview Media Asset

media_assets_preview
读取

Return one bounded JPEG as MCP image content for visual inspection. Images use the stored visual; videos use the existing thumbnail.

List Media Folders

media_folders_list
读取

List one bounded level of media-library folders and direct asset counts.

Create Media Folder

media_folders_create
写下

Create one media-library folder in a workspace and return a retry-safe mutation receipt.

Update Media Asset

media_assets_update
写下

Rename one media asset, edit its description, move it to a folder, or change B-roll eligibility. Provide at least one of name, description, folderId, or allowUsingAsBroll.

Estimate Video Upscale

media_upscale_estimate
读取

Estimate ByteDance upscale eligibility and credits for one stored video. Does not spend credits or queue a job. Call this before media_upscale.

Upscale Video

media_upscale
写下

Queue a ByteDance upscale of one stored video asset to 1080p or 4k. Call media_upscale_estimate first. Spends AI credits and returns a new pending media asset to poll with media_assets_get.

剪辑和字幕

24

Create Clips

clips_create
写下

Create a private BlitzReels clip batch from an existing asset or a user-provided video URL. Supports public social videos, Google Drive, and direct media URLs.

Get Clip Batch

clips_get
读取

Get status, generated clips, render counts, and download links for a BlitzReels clip batch.

Manage Clips

clips_manage
写下

List or inspect clips, regenerate, repair visual QA, start export, promote an accepted delivery export, inspect caption words, or apply a verified caption look.

Delete Clip Batch

clip_batches_delete
写下

Preview or delete a terminal clip batch. Dry-run is the default; real deletion requires explicit confirmation, retention choice, workspace delete permission, and an idempotency key.

Refresh Clip Progress

clips_status_get
读取

Read one existing batch's progress, ready clips and renewed signed links without opening or remounting a review widget. Safe for bounded polling and widget refresh; does not recreate or repair clips.

List Clips

clips_list
读取

List bounded managed clips from the selected source or project with their current workflow state.

Inspect One Clip

clips_inspect
读取

Read one managed clip's selected source range, project, workflow state and delivery information.

Revise Selected Moment

clips_reselect
写下

Replace the source moment for one managed clip using a verified suggestion or explicit time range. Can spend existing account credits. Preserves the other clips; read current state before retrying.

Repair One Clip Layout

clips_repair
写下

Request a layout repair for one managed clip, such as preserving a screen demo or moving captions away from faces. Changes that clip's project; preserves the other clips.

Export One Clip

clips_export
写下

Render one managed clip at an existing-account-supported quality. Returns persisted export state; inspect until the accepted delivery file is ready. Can spend existing account credits.

Apply Caption Look

captions_look_apply
写下

Apply a verified caption look to one project. Preserves word overrides unless clearWordOverrides=true. Existing captions are restyled; spoken audio is preserved.

List Synchronized Recordings

recording_sets_list
读取

Read bounded camera and screen recording sets in the selected workspace with status, duration and alignment warnings.

Inspect Synchronized Recording

recording_sets_get
读取

Read one camera and screen recording set, its primary speaker, content windows, alignment and warnings. Separate audio-only tracks are not supported.

Create Synchronized Recording

recording_sets_create
写下

Create an empty camera and screen recording set in the explicit workspace. Requires existing recording-set access; does not import files or start processing.

Attach Camera or Screen Recording

recording_sets_track_attach
写下

Attach one existing authorized video asset as a camera or screen track. Specify the main speaker when known; alignment happens during finalization.

Synchronize Recording Tracks

recording_sets_finalize
写下

Finalize the collected camera and screen tracks and queue analysis. Can spend existing transcription credits. Inspect returned alignment and warnings before clipping.

Clip a Synchronized Recording

recording_sets_clips_create
写下

Produce a batch from a ready synchronized camera and screen recording, with optional clip count, maximum duration and caption look. Preserves readable screen context through the existing layout pipeline. Can spend existing credits; follows clips_get with the returned batch_id.

Read Clip Review Link

clip_share_get
读取

Read whether the source video's clip share page is public and which clips are visible. Does not publish content.

Publish a Clip Review Link

clip_share_publish
写下

Make the selected source video's clip page public with reviewed title, description and explicit download permission. Requires confirmPublicSharing=true from the user. Returns the verified public URL; no social network post is made.

Unpublish a Clip Review Link

clip_share_unpublish
写下

Disable public access to the selected clip page. Changes an existing publication; preserves private projects and exports.

Set Shared Clip Visibility

clip_share_visibility_set
写下

Show or hide one identified clip on an existing public review page after explicit user confirmation. Preserves the other clips and private project data.

Create Project

projects_create
写下

Create a new private BlitzReels video project.

List Clip Presets

clips_presets_list
读取

List available clip batch presets and their clipping/caption defaults.

List Caption Themes

captions_themes_list
读取

List built-in caption looks and saved caption themes available to a workspace.

时间线编辑

11

Apply Timeline Edit

timeline_edit_apply
写下

Trim or extend one timeline item. Split one timeline item at an exact sequence position.

Preview Timeline Edit

timeline_edit_preview
读取

Preview without mutation. Trim or extend one timeline item. Split one timeline item at an exact sequence position.

Apply visual presets

visuals_apply
写下

Create or update up to 20 visual blocks in one call after edit timing is settled. Presets: broll-headline, steps, quote, end-screen, key-number, before-after, screenshot-focus, chapter. IDs are scoped to the project; repeating an ID updates it. Asset IDs must belong to the workspace. Atomic, undoable; dryRun validates without writing. Returns IDs and revision. No generation pass.

Update Timeline Clip

update_timeline_clip
写下

Update one timeline item's start, duration, trim, layer, ignored state, or blurred background. Provide at least one field to update besides timelineItemId.

Batch Update Timeline Clips

batch_update_timeline_clips
写下

Update timing, trim, layer, or state for up to 100 timeline items in one sequence transaction.

Mute Clip Audio

mute_clip_audio
写下

Set one timeline media item's volume to zero.

Delete Timeline Items

delete_timeline_items
写下

Preview or delete up to 100 timeline items and optionally remove linked captions.

Add Title Card

add_text_overlay
写下

Add an opening title card (hook) or generic text overlay to a project timeline. Defaults to classic white: black text on a white card.

Add Transition

add_transition
写下

Add a timed transition effect, optionally attached to one timeline item and paired with sound.

Add Media To Timeline

timeline_media_add
写下

Place an image or video asset onto a BlitzReels project timeline at a timestamp, including B-roll and static images.

Add Audio To Timeline

timeline_audio_add
写下

Place an audio asset onto a BlitzReels project timeline at a timestamp for voiceover, music, or sound.

工作流

07

List Workflows

workflows_list
读取

List public BlitzReels workflows that can be imported into a workspace.

Import Workflow

workflows_import
写下

Import a public BlitzReels workflow into the user's workspace.

Get Workflow

workflows_get
读取

Get a workflow definition, input schema, defaults, and ordered steps.

Create Workflow Run

workflow_runs_create
写下

Create an interactive workflow run from a workflow and optional inputs.

Get Workflow Run

workflow_runs_get
读取

Get workflow run status, step statuses, outputs, warnings, and dashboard URL.

Execute Workflow Run Step

workflow_runs_execute_step
写下

Execute one pending workflow run step and return updated run status.

Start Export

exports_start
写下

Render a BlitzReels video project and return export and job IDs for status checks.

AI 生成

17

List Generation Options

generation_options_list
读取

List the models, defaults, limits, and credit costs available for one generation kind before queueing it. Video duration is an enum per model (durations_seconds), not a min-max range.

Generate Image

generation_image_create
写下

Queue an AI image generation into the BlitzReels media library. Spends AI credits and returns a job to poll with generation_jobs_get.

Generate Video

generation_video_create
写下

Queue video generation, image animation, source-video editing or character replacement into the media library. Editing accepts referenceVideoAssetIds and transformation controls, including original-audio preservation. Spends AI credits and returns a job readable with generation_jobs_get.

Generate Voiceover

generation_voiceover_create
写下

Queue a text-to-speech voiceover into the BlitzReels media library. Spends AI credits and returns a job to poll with generation_jobs_get.

Generate Music

generation_music_create
写下

Queue a background music track into the BlitzReels media library. Spends AI credits and returns a job to poll with generation_jobs_get.

Generate Sound Effect

generation_sound_create
写下

Queue a sound effect into the BlitzReels media library. Spends AI credits and returns a job to poll with generation_jobs_get.

Generate Faceless Video

generation_faceless_create
写下

Turn a script into a full faceless video project: scene plan, generated visuals, optional voiceover, captions, music, and sound. Spends AI credits and returns a job to poll with generation_jobs_get.

Get Generation Job

generation_jobs_get
读取

Check a queued generation. Completed jobs carry the media asset ID and a download URL.

Cancel Generation Job

generation_jobs_cancel
写下

Request cancellation of the identified queued or running generation. Work already performed can consume credits; cancellation does not establish a refund.

Get Generation Plan Brief

generation_plan_brief
读取

Read a generation brief from rawPrompt. Source-video edits return execution for generation_video_create with required assets and transformation controls. New scenes return an authoring grammar and planJsonSchema for generation_plan_validate. This read does not generate media or spend credits.

Validate Generation Plan

generation_plan_validate
读取

Validate a self-authored generation plan against the schema and the target model's real capabilities before execution. Returns directive errors to fix and revalidate; only a valid plan is executable.

Plan an Approved Video Script

video_storyboard_create
写下

Create a vertical video storyboard for the user-approved script in an existing project. Spends planning credits; returns scene narration, factual review notes, references and the production credit estimate. Does not generate images or animate video. Existing account access applies.

Review Video Storyboard

video_storyboard_get
读取

Read one authorized storyboard revision, its scene and reference previews, estimate, approvals and bounded polling guidance. Refreshes signed previews. Complete means the editable timeline is assembled; use exports_start and exports_list for a downloadable MP4.

Revise Selected Storyboard Scenes

video_storyboard_revise
写下

Create a new immutable draft from a known revision with selected scene, character or model changes. Preserves other scene definitions; requires review and production approval again. Does not generate media or overwrite an approved revision.

Approve Storyboard Previews

video_storyboard_approve
写下

After showing the storyboard and its credit estimate, confirmApproved=true starts production of reference images and scene keyframes. Spends existing credits. This stage produces reviewable previews, not animated videos.

Approve Previews and Produce Video

video_assets_approve
写下

After the user approves the generated previews and animation cost, confirmApproved=true starts scene animation, narration and editable timeline assembly. Can replace the project's generated timeline and spend existing credits. Follow the revision and then export the finished project.

Regenerate Selected Storyboard Preview

video_storyboard_asset_regenerate
写下

Replace the identified preview after user approval of the new credit cost. A scene_keyframe changes one keyframe; a character_anchor refreshes its dependent keyframes; a style_anchor refreshes every keyframe. Preserves unaffected previews. Does not regenerate an already finished video shot.

故事套件

06

List Reusable Characters

characters_list
读取

List bounded reusable characters with identity, voice, facts, and image-reference metadata.

Get Reusable Character

characters_get
读取

Inspect one reusable character and its ordered reference-image set before planning a story.

Apply Reusable Character

characters_apply
写下

Create or update a reference-based character for repeatable faceless story generation.

List Story Kits

story_kits_list
读取

List bounded reusable Story Kits that bind characters, locations, style, facts, and narrator voice.

Get Story Kit

story_kits_get
读取

Inspect one Story Kit and its revision before generating or updating a story.

Apply Story Kit

story_kits_apply
写下

Create or update a Story Kit with optimistic concurrency and an idempotent mutation receipt.

系列

06

List Series

series_list
读取

List optional Series with bounded counts and cover previews.

Get Series

series_get
读取

Read a Series identity, creative defaults and revision.

List Series Content

series_content
读取

List sources or projects in a Series. Null seriesId lists standalone content. sourceId filters projects by clip origin.

Apply Series

series_apply
写下

Create or replace a Growth+ Series identity and defaults. defaultCreationMode selects podcast branding or faceless Story Kit defaults; membership stays mixed. Updates require expectedRevision. Existing content keeps its settings.

Assign Series Content

series_assign
写下

Assign, move or detach one source or project using expectedSeriesId. Moving a source does not move existing clips or change branding.

Delete Series

series_delete
写下

Delete the Series container at an expected revision and detach its content. Sources, projects and exports are preserved.

轮播图文

09

Compile Carousel

carousels_compile
写下

Create or replace a carousel project from a slide document. Server owns chip layout, safezones, and exact export pixels.

Preview Carousel

carousels_preview
读取

Render JPEG stills for a compiled carousel at the exact platform frame and return them as MCP images so you can inspect chips on the real slides.

Get Carousel Document

carousels_get
读取

Read the carousel slot document and the named template catalog.

Patch Carousel Slide

carousels_patch_slide
写下

Patch named slots or template_id on one slide. Re-rasters that slide only. Never rewrite HTML.

Patch Carousel Document

carousels_patch_document
写下

Reorder, remove, insert, duplicate, or undo slides on a compiled carousel. One write for the filmstrip. Insert rasters an empty named template unless sourceIndex copies a slide.

Expand Carousel Recipe

carousels_expand_recipe
读取

Turn a named recipe or a brief into compile-ready named-template slides. Expand never publishes and never invents metrics.

Migrate Legacy Carousel

carousels_migrate
写下

Rebuild a compile slot document from a legacy overlay carousel. Then patch slots like any compiled deck.

Clone Carousel

carousels_clone
写下

Duplicate a carousel onto a new project. Copies slots. Icon and shot values may be pack filenames or asset:<media_asset_id>.

Carousel Theme Kit

carousels_kit
写下

Save or apply a theme, brand theme, or named slide template. Themes rewrite fonts, colors, overlay, and asset:<id> logos. Never persist signed URLs.

发布

05

List Publishing Connections

publishing_connections_list
读取

List Instagram accounts connected through the BlitzReels UI for the authenticated workspace, including their connection status.

Schedule Instagram Publish

publishing_instagram_schedule
写下

Schedule a completed project export to Instagram. Optional includeAffiliateCta appends Try BlitzReels for free plus the Affonso tracking URL. Requires an active Instagram account connected in the BlitzReels UI.

Read Connected Publishing Accounts

publishing_accounts_list
读取

Read the supported accounts available through the existing workspace and provider access rules. Does not connect new accounts, publish, schedule or bypass provider review.

Schedule Reviewed Video Publication

publishing_video_schedule
写下

Schedule a selected existing video asset to explicit active YouTube, Instagram, Facebook, TikTok, LinkedIn or Threads accounts connected in the BlitzReels UI, using reviewed copy, privacy/consent choices, synthetic-media disclosure, time and timezone. Requires confirmPublish=true. Existing workspace permissions and provider access rules apply. A scheduled receipt is not proof of publication.

Verify Video Publication

publishing_post_get
读取

Read the current targets for one publishing receipt with scheduled/published state and verified remote URLs. Does not retry or publish a failed target.

协议优先

无首选模型。无私有集成。

BlitzReels通过Streamable HTTP实现MCP。任何支持该协议和OAuth的客户端都可使用相同工具。

Claude
Codex
ChatGPT
Cursor
Gemini
OpenClaw

AI代理视频编辑常见问题

关于托管服务器的连接、认证、兼容性和API问题。

BlitzReels有MCP服务器吗?+

有。BlitzReels通过Streamable HTTP和OAuth运行托管MCP服务器,地址为https://blitzreels.com/api/mcp。添加到ChatGPT、Claude、Cursor或任何支持MCP的客户端,代理即可代表您读取项目、编辑和导出。

Claude Code能通过CLI编辑带字幕的视频吗?+

可以。Claude Code通过BlitzReels CLI(npx blitzreels auth login)登录,然后从终端读取项目、编辑字幕和转录、插入B-roll并导出。可安装的代理技能包提供精确命令和安全编辑流程。

哪些AI代理兼容BlitzReels?+

BlitzReels支持Claude Code、OpenAI Codex和ChatGPT通过MCP服务器、CLI或REST API,以及其他支持MCP的客户端如Cursor。常用编码代理有可安装的技能包。

哪些视频API支持转录、B-roll和导出自动化?+

BlitzReels在一个REST API中涵盖转录、字幕、B-roll插入、重构和导出自动化,配有托管MCP服务器和CLI。/api/openapi.json的OpenAPI模式支持为自定义代理生成客户端。

我可以免费试用代理集成吗?+

可以。通过网站注册后可进入为期 7 天、需要银行卡的试用期;试用期导出带水印。无水印导出需要有效的付费订阅。

给您的代理一个视频编辑器。

连接一次,批准工作区,即可从任何支持MCP的客户端开始编辑。