@genspark/cli 1.15.1 → 1.15.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +14 -1
- package/dist/client.d.ts +8 -0
- package/dist/client.d.ts.map +1 -1
- package/dist/client.js +169 -22
- package/dist/client.js.map +1 -1
- package/dist/commands/aidrive-contract.generated.d.ts.map +1 -1
- package/dist/commands/aidrive-contract.generated.js +48 -4
- package/dist/commands/aidrive-contract.generated.js.map +1 -1
- package/dist/commands/aidrive-operands.d.ts.map +1 -1
- package/dist/commands/aidrive-operands.js +9 -0
- package/dist/commands/aidrive-operands.js.map +1 -1
- package/dist/commands/aidrive-upload.js +2 -2
- package/dist/commands/aidrive-upload.js.map +1 -1
- package/dist/config.d.ts +9 -1
- package/dist/config.d.ts.map +1 -1
- package/dist/config.js +29 -3
- package/dist/config.js.map +1 -1
- package/dist/index.js +5 -3
- package/dist/index.js.map +1 -1
- package/dist/types.d.ts +4 -0
- package/dist/types.d.ts.map +1 -1
- package/package.json +1 -1
- package/skills/gsk-aidrive/SKILL.md +7 -2
- package/skills/gsk-create-task/SKILL.md +1 -1
- package/skills/gsk-genteam/SKILL.md +1 -1
- package/skills/gsk-google-calendar/SKILL.md +1 -1
- package/skills/gsk-outlook-calendar/SKILL.md +1 -1
- package/skills/gsk-second-brain/SKILL.md +1 -1
- package/skills/gsk-shared/SKILL.md +22 -0
- package/skills/gsk-video-generation/SKILL.md +1 -1
package/package.json
CHANGED
|
@@ -403,7 +403,7 @@ in the UI.
|
|
|
403
403
|
| `--task_tags_remove` | No | For task_update: remove tag names case-insensitively; missing names are a no-op. Cannot also add the same name or replace/clear tags. (array) |
|
|
404
404
|
| `--task_filter_tags` | No | For task_list: require ALL exact tag names (case-insensitive). Combines with status/assignee/priority filters using AND; empty means no tag filter. (array) |
|
|
405
405
|
| `--ref` | No | For link into an SB Workspace: Genspark project UUID or project URL, agentbase:<database UUID>, or meeting:<meeting UUID>. Use path instead for an original My Drive or Team Drive file/folder; source selects the destination Workspace. (string) |
|
|
406
|
-
| `<action>` (positional) | Yes | Action to perform. create: Create an SB Workspace with its own Drive; requires workspace=sb_workspace. ls: List a directory, or discover shared sources when no source is selected. find: Find files by name keyword; scope selects mine, shared or all. mkdir: Create a directory in the selected Drive. rm: Move the selected entry to recoverable trash; removing a shortcut preserves its source. With workspace=sb_workspace and only source, delete the entire Workspace (owner only). move: Move or rename within one Drive source; never overwrite an existing file. get_readable_url: Get temporary file bytes URL and the separate document URL for people. download_video: Import an internet video into Drive (YouTube is billed by delivered size). download_audio: Import internet audio into Drive (YouTube is billed by delivered size). download_file: Import an external file URL into Drive; local downloads use download. compress: Create an archive from a Drive folder. decompress: Extract an archive in Drive. share: Grant access to an entry; recipient and general-access forms use existing sharing rules. unshare: Revoke an entry's recipient or general access. access: Inspect who can open the selected entry. link: Place a shortcut in a managed Drive. For an SB Workspace, use path for an original My Drive/Team Drive entry or ref for a Genspark project, meeting, or AgentBase resource. unlink: Remove the selected shortcut and its grant without deleting source bytes. upload: Upload a local file or inline content into Drive; local and inline conflict policies differ. list_trash: List recoverable entries in My Drive trash. restore: Restore a trash entry by its stable entry_id, optionally into another folder. tree: Browse bounded directory structure; continue with the returned cursor. search: Search names and indexed content; continue with the returned cursor. read: Read persisted extracted text by locator, batch ids, or a returned read cursor. overview: Read cached summaries; missing summaries prepare asynchronously without blocking reads. copy: Copy entries into a writable Drive, or inspect/continue an existing job with job_id. info: Read an SB Workspace's shared context: brief, team instructions, your access, memo, open tasks and the commands that work in it. task_list: List Workspace tasks; status, assignee, priority and all requested tags combine as AND filters. task_add: Add an SB Workspace task with optional status, due date, assignee, links, priority, details and tags. task_update: Change an SB Workspace task's fields; omitted or blank values leave fields unchanged. task_done: Mark an SB Workspace task done; task_update --task_status open reopens it. task_rm: Delete an SB Workspace task. grep: Find literal phrases in one file; continue with the returned cursor. (string, one of: create, ls, find, mkdir, rm, move, get_readable_url, download, download_video, download_audio, download_file, compress, decompress, share, unshare, access, link, unlink, upload, list_trash, restore, tree, search, read, overview, copy, info, task_list, task_add, task_update, task_done, task_rm, grep) |
|
|
406
|
+
| `<action>` (positional) | Yes | Action to perform. create: Create an SB Workspace with its own Drive; requires workspace=sb_workspace. ls: List a directory, or discover shared sources when no source is selected. find: Find files by name keyword; scope selects mine, shared or all. tag_add: Add shared file/folder tags without changing other tags. References keep local tags. tag_rm: Remove shared tags by case-insensitive name, preserving other tags. mkdir: Create a directory in the selected Drive. rm: Move the selected entry to recoverable trash; removing a shortcut preserves its source. With workspace=sb_workspace and only source, delete the entire Workspace (owner only). move: Move or rename within one Drive source; never overwrite an existing file. get_readable_url: Get temporary file bytes URL and the separate document URL for people. download_video: Import an internet video into Drive (YouTube is billed by delivered size). download_audio: Import internet audio into Drive (YouTube is billed by delivered size). download_file: Import an external file URL into Drive; local downloads use download. compress: Create an archive from a Drive folder. decompress: Extract an archive in Drive. share: Grant access to an entry; recipient and general-access forms use existing sharing rules. unshare: Revoke an entry's recipient or general access. access: Inspect who can open the selected entry. link: Place a shortcut in a managed Drive. For an SB Workspace, use path for an original My Drive/Team Drive entry or ref for a Genspark project, meeting, or AgentBase resource. unlink: Remove the selected shortcut and its grant without deleting source bytes. upload: Upload a local file or inline content into Drive; local and inline conflict policies differ. list_trash: List recoverable entries in My Drive trash. restore: Restore a trash entry by its stable entry_id, optionally into another folder. tree: Browse bounded directory structure; continue with the returned cursor. search: Search names and indexed content; continue with the returned cursor. read: Read persisted extracted text by locator, batch ids, or a returned read cursor. overview: Read cached summaries; missing summaries prepare asynchronously without blocking reads. copy: Copy entries into a writable Drive, or inspect/continue an existing job with job_id. info: Read an SB Workspace's shared context: brief, team instructions, your access, memo, open tasks and the commands that work in it. task_list: List Workspace tasks; status, assignee, priority and all requested tags combine as AND filters. task_add: Add an SB Workspace task with optional status, due date, assignee, links, priority, details and tags. task_update: Change an SB Workspace task's fields; omitted or blank values leave fields unchanged. task_done: Mark an SB Workspace task done; task_update --task_status open reopens it. task_rm: Delete an SB Workspace task. grep: Find literal phrases in one file; continue with the returned cursor. (string, one of: create, ls, find, tag_add, tag_rm, mkdir, rm, move, get_readable_url, download, download_video, download_audio, download_file, compress, decompress, share, unshare, access, link, unlink, upload, list_trash, restore, tree, search, read, overview, copy, info, task_list, task_add, task_update, task_done, task_rm, grep) |
|
|
407
407
|
| `--workspace` | No | AI Drive location. my_drive owns your bytes and quota; shared_drive selects all managed Team, GenTeam, Hub, SB Workspace, and authorized knowledge drives; sb_workspace selects only Second Brain Workspace drives; team_drive selects only Team Drives; genteam selects only GenTeam channel and DM drives. shared_with_me is an ACL-filtered view across both kinds of container and owns no bytes or quota. GenTeam drives are flat: upload files or share source folders at the root; no native mkdir. In GenTeam channel and DM drives, moving, renaming, and deleting native files needs a channel admin; Team Drive editors can do all three. (string, one of: my_drive, shared_with_me, shared_drive, sb_workspace, team_drive, genteam, default: `my_drive`) |
|
|
408
408
|
| `--source` | No | Stable source_id returned by ls in a shared workspace. It identifies the storage-owning drive and entry; names are accepted only when unique. For My Drive use workspace=my_drive and omit source; my_drive and Second Brain memo repo ids are not source_ids. (string) |
|
|
409
409
|
| `--share` | No | Deprecated alias for source, retained for older gsk scripts. (string) |
|
|
@@ -415,7 +415,11 @@ in the UI.
|
|
|
415
415
|
| `--general_access` | No | Who else can open it, for share without a recipient: restricted, org, or link. Also accepted by upload into My Drive, to open the new file at once. (string, one of: restricted, org, link) |
|
|
416
416
|
| `--organization_id` | No | Which organization to share with, when you belong to more than one and use to=org or to=group:<uid>. (string) |
|
|
417
417
|
| `--expires_at` | No | Optional ISO-8601 UTC instant after which the access stops, e.g. 2026-12-31T00:00:00+00:00. (string) |
|
|
418
|
-
| `-p`, `--path` | No | Path to file or folder for ls, mkdir, rm, move, get_readable_url. For link: the existing source path in My Drive. For unlink: the LinkNode path in the selected managed source. For compress: folder path to compress. For decompress: archive file path to extract. For find: which directory to search. Omit it to search the whole of My Drive, or the root of the selected shared source.
|
|
418
|
+
| `-p`, `--path` | No | Path to file or folder for ls, mkdir, rm, move, get_readable_url. For link: the existing source path in My Drive. For unlink: the LinkNode path in the selected managed source. For compress: folder path to compress. For decompress: archive file path to extract. For find: which directory to search. Omit it to search the whole of My Drive, or the root of the selected shared source. With tags or untagged, find searches the selected subtree. Without tag filters, naming a directory searches THAT DIRECTORY ONLY and does not descend into its subfolders — and a shared folder's own root is one such directory. To cover a subtree, use search with query_scope=subtree. tree/search/read/overview also accept a path relative to source or id. With a file id/entry_id, its full path selects that same file. With a directory id/entry_id, paths resolve beneath it. For a known file, id/entry_id alone is sufficient. (string) |
|
|
419
|
+
| `--tags` | No | Shared file tag names for tag_add, tag_rm or find. (array) |
|
|
420
|
+
| `--tag_color` | No | Color for added file tags; defaults to blue. (string, one of: gray, red, orange, yellow, green, blue, purple, pink) |
|
|
421
|
+
| `--tag_match` | No | find: match any selected tag (default), or all. (string, one of: any, all) |
|
|
422
|
+
| `--untagged` | No | find: include entries without tags. (boolean) |
|
|
419
423
|
| `-q`, `--query` | No | Find: words in names/link metadata. Search: words in names and indexed content (AND). Grep: a literal phrase in one file. Read: a literal phrase anchoring the first matching passage. Phrase matching ignores case and whitespace differences. (string) |
|
|
420
424
|
| `-f`, `--filter_type` | No | Filter by entry type for ls (improves performance): all (default), file, directory. Use 'file' when only need files. (string, one of: all, file, directory) |
|
|
421
425
|
| `--file_type` | No | Filter by file MIME type for ls (improves performance): all (default), audio, video, image. Combine with filter_type='file' for best results. (string, one of: all, audio, video, image) |
|
|
@@ -473,6 +477,7 @@ Use the canonical actions below. `gsk drive <action> --help` shows the operation
|
|
|
473
477
|
- Only when the source is unknown, use `ls` for My Drive, `ls --workspace sb_workspace` for SB Workspaces, `ls --workspace team_drive` for Team Drives, `ls --workspace genteam` for GenTeam channel and DM drives, `ls --workspace shared_drive` for all managed sources, or `ls --workspace shared_with_me` for all shared sources. Reuse source IDs; duplicate source names require an explicit ID.
|
|
474
478
|
- With only a name, use `find --scope all -q <name>`; `mine` and `shared` narrow it. Reuse a hit's `entry_locator` (workspace/source/path). Search is bounded and may use an incomplete index: inspect coverage and never interpret no matches as proof of absence.
|
|
475
479
|
- A LinkNode path follows its live source for reads, folder browsing and uploads. Its `content_source_id`/`link_workspace_id` also opens that source at `/`. Do not combine the content mount with the alias's old path. `rm`/`move` on the alias operate on the alias; on a child they operate inside its source. Cross-source moves are refused. Shared source roots cannot be removed. Source-owned shared projections cannot be moved/renamed; unlinking may end that audience's share, reported by share_ended, while preserving source bytes.
|
|
480
|
+
- A `web` link (`link_target_kind: web`) has no bytes and no source to follow: read the page with `gsk crawl <link_target_url>` (the SSRF-guarded crawler); never fetch it with curl or wget.
|
|
476
481
|
- `get_readable_url` returns the native `entry_id` and, for Office files, `collaboration.file_id` and `link_workspace_id`. For live editing use `wo-peer open <file_id> --link-workspace-id <context>` (omit the flag for My Drive). Never use the shortcut ID as the room ID. Minting rechecks edit access; PDF and project links are not Office rooms.
|
|
477
482
|
- A DM reference grants no source access by itself. Revoked/unavailable/view-only results are permission outcomes; retrying download on the LinkNode cannot fix them.
|
|
478
483
|
|
|
@@ -32,7 +32,7 @@ gsk task create [options]
|
|
|
32
32
|
|
|
33
33
|
| Flag | Required | Description |
|
|
34
34
|
|------|----------|-------------|
|
|
35
|
-
| `<task_type>` (positional) | Yes | The type of task to create: - super_agent: Create general super agent - podcasts: Create audio podcasts with AI characters - docs: Create and edit HTML/Markdown documents. Create a professional document with Word-like formatting as a single-file HTML. Perfect for reports, articles, academic papers, and formal documents. **Guidelines** 1. This tool can be directly called when user explicitly wants to create: - A document or report - An article or academic paper - A formal document with professional formatting - A document for print or PDF export 2. In situations other than those mentioned above, you need to ask the user first and only call this tool after getting permission when user wants to create: - A better formatted text, but hasn't explicitly mentioned creating a document - Content that needs professional document formatting **Limitations** - Document styling resembles Microsoft Word format, not web page design - Usually takes time to generate, so please strictly follow the guidelines above **File inputs** - If the user uploaded a source file (PDF/Word/Markdown/text) for the agent to process or convert, embed the URL (file_wrapper_url, http(s)://, or aidrive://) directly in the query — sub-agents do not inherit parent context. - slides: Create presentation slides. Create an editable slide deck and display it; it exports/downloads as PPTX or PDF. DEFAULT for any presentation, deck, or PowerPoint request — including a downloadable .pptx or redesigning an uploaded .pptx; do NOT rebuild decks in code or raw HTML. EXCEPTION — simple mechanical edits to an uploaded .pptx (insert/delete a slide, change a value, swap an element/image): do them yourself with python-pptx on the original file and deliver the edited file; this agent re-authors the deck and its export is a re-render, so untouched slides won't stay byte-identical. Route substantial rework (redesign, restyle, use as a template, content overhaul) to this agent. The finished deck's page (task_url) has built-in PPTX/PDF export — NEVER create another task to export or convert an existing deck; use task_status to re-fetch a deck's link. Limitation: a) It usually takes minutes. Ask the user whether they want to wait for a longer time before starting if they haven't explicitly indicated they want to generate a set of slides; This slides agent maintains its own internal library of presentation skills (templates, narrative patterns, visual styles) that you do not have visibility into. When the user references a skill, technique, or style by name that you do not recognize as one of your own skills, do NOT try to interpret, expand, or rewrite it. If the user gave an exact skill slug (one they explicitly named or picked), pass that slug in the `skill_name` parameter (or `public_skill_name` for a public-gallery skill) so the slides agent loads it directly; otherwise forward the user's wording verbatim in the query so the slides agent can look it up against its own skill library. Likewise, pass through any slide-specific stylistic guidance from the user without paraphrasing it. It authors `<deck>.slides/` decks; the user's preview and the built-in PPTX/PDF export read ONLY those decks. Surgically editing a native .pptx in place is NOT supported here — a script-built .pptx (python-pptx etc.) looks crude, and lands as a loose file the preview cannot render and export cannot deliver, so the user gets nothing to download (simple mechanical edits belong in your own python-pptx pass, per the EXCEPTION above). When you do dispatch rework of an uploaded .pptx, never brief 'edit the original file directly' or 'deliver a hand-built .pptx'; brief this flow: replicate the upload as a .slides deck matching its design as closely as possible, apply the requested changes, deliver via the built-in export (a re-render, not byte-identical — set that expectation with the user). - website: Create a professional agent for building websites and web pages - deep_research: Create deep research agent for task on a 'deep research' topic (e.g. 'deep research on Tesla', 'analyze Tesla in depth'). Also can handle user's 'continue deep research' request - download: Download a file, video, or audio from the internet into the user's AI Drive and return a long-lived shareable link. Use when the user wants to 'download this', 'save this video/file', or 'get me a link to this'. The agent locates the real downloadable resource (searching/crawling a page if needed), downloads it, and returns the AI Drive path plus a durable share URL. - design: Genspark Design (Designer 2.0): the persistent visual web/HTML design canvas for landing pages, marketing sites, posters, slides-as-pages, and interactive UI mockups. It pre-creates a stable project_id and hands the user its Web canvas with query and instructions combined into the single visible prompt field (the user submits it there). The same project can later be inspected with task_status and edited with task_continue. Use when the user wants a web page / site / poster / visual layout they will iterate on in the Design canvas. - video_generation: Large-scale batch video generation with complex inter-task dependencies. Best for workflows requiring coordinated, multi-step video pipelines where tasks depend on each other's outputs. Supports text-to-video, image-to-video (query should contain the image url), and merging multiple video clips. Can reuse agent. NOT recommended for small-scale, simple, or independent video tasks — for those, use <tool>video_generation</tool> directly instead. If there is a resource generation list, you can directly pass the resource list URL in the query, http(s):// or aidrive:// - audio_generation: Large-scale batch TTS (text-to-speech) generation with complex inter-task dependencies. Best for workflows requiring coordinated, multi-step TTS pipelines where tasks depend on each other's outputs. Supports text-to-speech, voiceovers, multi-speaker dialogues, and merging multiple audio clips. Can reuse agent. NOT recommended for small-scale, simple, or independent TTS tasks — for those, use <tool>audio_generation</tool> directly instead. If there is a resource generation list, you can directly pass the resource list URL in the query, http(s):// or aidrive:// - cross_check: Fact-check and verify claims or statements. Use when the user wants to verify whether a statement is true, check facts, or cross-reference claims against multiple sources. The agent systematically searches for evidence, evaluates source credibility, and provides a detailed verification report with supporting references. - meeting_notes - sheets - custom_super_agent: Execute a user-created custom AI agent. Requires agent_config_id parameter which should be extracted from the workflow step data. These are personalized agents created by users with specific capabilities and configurations. (string, one of: super_agent, podcasts, docs, slides, website, deep_research, download, design, video_generation, audio_generation, cross_check, meeting_notes, sheets, custom_super_agent) |
|
|
35
|
+
| `<task_type>` (positional) | Yes | The type of task to create: - super_agent: Create general super agent - podcasts: Create audio podcasts with AI characters - docs: Create and edit HTML/Markdown documents. Create a professional document with Word-like formatting as a single-file HTML. Perfect for reports, articles, academic papers, and formal documents. **Guidelines** 1. This tool can be directly called when user explicitly wants to create: - A document or report - An article or academic paper - A formal document with professional formatting - A document for print or PDF export 2. In situations other than those mentioned above, you need to ask the user first and only call this tool after getting permission when user wants to create: - A better formatted text, but hasn't explicitly mentioned creating a document - Content that needs professional document formatting **Limitations** - Document styling resembles Microsoft Word format, not web page design - Usually takes time to generate, so please strictly follow the guidelines above **File inputs** - If the user uploaded a source file (PDF/Word/Markdown/text) for the agent to process or convert, embed the URL (file_wrapper_url, http(s)://, or aidrive://) directly in the query — sub-agents do not inherit parent context. - slides: Create presentation slides. Create an editable slide deck and display it; it exports/downloads as PPTX or PDF. DEFAULT for any presentation, deck, or PowerPoint request — including a downloadable .pptx or redesigning an uploaded .pptx; do NOT rebuild decks in code or raw HTML. EXCEPTION — simple mechanical edits to an uploaded .pptx (insert/delete a slide, change a value, swap an element/image): do them yourself with python-pptx on the original file and deliver the edited file; this agent re-authors the deck and its export is a re-render, so untouched slides won't stay byte-identical. Route substantial rework (redesign, restyle, use as a template, content overhaul) to this agent. The finished deck's page (task_url) has built-in PPTX/PDF export — NEVER create another task to export or convert an existing deck; use task_status to re-fetch a deck's link. Limitation: a) It usually takes minutes. Ask the user whether they want to wait for a longer time before starting if they haven't explicitly indicated they want to generate a set of slides; This slides agent maintains its own internal library of presentation skills (templates, narrative patterns, visual styles) that you do not have visibility into. When the user references a skill, technique, or style by name that you do not recognize as one of your own skills, do NOT try to interpret, expand, or rewrite it. If the user gave an exact skill slug (one they explicitly named or picked), pass that slug in the `skill_name` parameter (or `public_skill_name` for a public-gallery skill) so the slides agent loads it directly; otherwise forward the user's wording verbatim in the query so the slides agent can look it up against its own skill library. Likewise, pass through any slide-specific stylistic guidance from the user without paraphrasing it. It authors `<deck>.slides/` decks; the user's preview and the built-in PPTX/PDF export read ONLY those decks. Surgically editing a native .pptx in place is NOT supported here — a script-built .pptx (python-pptx etc.) looks crude, and lands as a loose file the preview cannot render and export cannot deliver, so the user gets nothing to download (simple mechanical edits belong in your own python-pptx pass, per the EXCEPTION above). When you do dispatch rework of an uploaded .pptx, never brief 'edit the original file directly' or 'deliver a hand-built .pptx'; brief this flow: replicate the upload as a .slides deck matching its design as closely as possible, apply the requested changes, deliver via the built-in export (a re-render, not byte-identical — set that expectation with the user). - website: Create a professional agent for building websites and web pages - deep_research: Create deep research agent for task on a 'deep research' topic (e.g. 'deep research on Tesla', 'analyze Tesla in depth'). Also can handle user's 'continue deep research' request - download: Download a file, video, or audio from the internet into the user's AI Drive and return a long-lived shareable link. Use when the user wants to 'download this', 'save this video/file', or 'get me a link to this'. The agent locates the real downloadable resource (searching/crawling a page if needed), downloads it, and returns the AI Drive path plus a durable share URL. - design: Genspark Design (Designer 2.0): the persistent visual web/HTML design canvas for landing pages, marketing sites, posters, slides-as-pages, and interactive UI mockups. It pre-creates a stable project_id and hands the user its Web canvas with query and instructions combined into the single visible prompt field (the user submits it there). The same project can later be inspected with task_status and edited with task_continue. Use when the user wants a web page / site / poster / visual layout they will iterate on in the Design canvas. - video_generation: Large-scale batch video generation with complex inter-task dependencies. Best for workflows requiring coordinated, multi-step video pipelines where tasks depend on each other's outputs. Supports text-to-video, image-to-video (query should contain the image url), and merging multiple video clips. Can reuse agent. NOT recommended for small-scale, simple, or independent video tasks — for those, use <tool>video_generation</tool> directly instead. If there is a resource generation list, you can directly pass the resource list URL in the query, http(s):// or aidrive:// - audio_generation: Large-scale batch TTS (text-to-speech) generation with complex inter-task dependencies. Best for workflows requiring coordinated, multi-step TTS pipelines where tasks depend on each other's outputs. Supports text-to-speech, voiceovers, multi-speaker dialogues, and merging multiple audio clips. Can reuse agent. NOT recommended for small-scale, simple, or independent TTS tasks — for those, use <tool>audio_generation</tool> directly instead. If there is a resource generation list, you can directly pass the resource list URL in the query, http(s):// or aidrive:// - cross_check: Fact-check and verify claims or statements. Use when the user wants to verify whether a statement is true, check facts, or cross-reference claims against multiple sources. The agent systematically searches for evidence, evaluates source credibility, and provides a detailed verification report with supporting references. - meeting_notes - sheets - custom_super_agent: Execute a user-created custom AI agent. Requires agent_config_id parameter which should be extracted from the workflow step data. These are personalized agents created by users with specific capabilities and configurations. - media_agent: Genspark Media agent: one production agent for images, video, audio/music and 3D on an infinite canvas with a per-project asset library. Runs to completion server-side with no user present — a clarifying-questions form or a require_approval draft ends the turn awaiting the user on the canvas, so put the full deliverable spec in query. The returned task_url opens the canvas, which attaches to the running turn (progress streams in; a message typed there joins the current turn) and stays the place to review the assets and continue; follow-ups from the CLI go through task_continue. (string, one of: super_agent, podcasts, docs, slides, website, deep_research, download, design, video_generation, audio_generation, cross_check, meeting_notes, sheets, custom_super_agent, media_agent) |
|
|
36
36
|
| `--task_name` | Yes | Name for the task/project (string) |
|
|
37
37
|
| `--query` | Yes | Query for the task. The ending should clearly state what needs to be done, such as generating a doc/slides etc. The sub-agent sees ONLY what is passed via query/instructions — it cannot read your conversation. When you have already produced material for this task (research notes, analysis, drafts, exact figures), embed its FULL text verbatim in the query, NOT a summary or paraphrase: any detail omitted here is missing from the deliverable, and a query carrying such material is naturally long — do not shorten it. For a simple request with no pre-existing material, a concise query is fine. If the user uploaded files or you have data the sub-agent needs, embed the file URLs (file_wrapper_url, http(s)://, or aidrive://) directly in the query — sub-agents do not inherit parent context, so unreferenced files are invisible to them. (string) |
|
|
38
38
|
| `--instructions` | Yes | Detailed instructions for the task. This will be set as system prompt to the specialized agent. State the working constraints here: audience, structure, style, output format, and what must be preserved exactly (figures, citations, links). Do not spend this field on a bare persona line like 'You are a professional slides assistant' — that gives the sub-agent nothing to work with. Include any file URLs (file_wrapper_url, http(s)://, or aidrive://) the sub-agent must reference; the sub-agent only sees what is passed via query/instructions. Design is the exception to the system-prompt transport: its Web canvas has one visible prompt field, so create_task appends instructions under an 'Additional requirements' heading in that field for the user to review before submitting. (string) |
|
|
@@ -53,7 +53,7 @@ gsk genteam [options]
|
|
|
53
53
|
| `--mentions` | No | [send] Members to @mention — humans or agents: agent ids (agent_…, from genteam_list_agents), human member ids (UUID), or exact display names (must be unambiguous in the target conversation). Each must already be a member of the target conversation. Mention chips render natively. Humans are notified; agents wake unless creator-only mode blocks the sender. Read any owner_only_mentions advisory in the reply before expecting an answer. A canonical @[name](id) token is auto-prepended to the body for each mention, skipped when the body already contains that member's token — so do NOT also write the token inline for the same member (it would be redundant, not doubled). (array) |
|
|
54
54
|
| `--quote_message_id` | No | [send] comet_message_id of an existing message in this conversation (from genteam_read) to quote — renders as the app's native quote block. Text sends only. (string) |
|
|
55
55
|
| `--operation_id` | No | [send] Optional idempotency key (8-64 chars of [A-Za-z0-9_-]). After a successful send, retries with the same operation_id return the already-sent message instead of posting a duplicate. A re-run of a send that was held for confirmation must carry the same operation_id and arguments to be recognised as the same send. \| [forward] Optional idempotency key (8-64 chars of [A-Za-z0-9_-]). After a successful forward, retries with the same operation_id return the already-forwarded message instead of posting a duplicate. Without it a re-run posts a duplicate forward. (string) |
|
|
56
|
-
| `--skip_sent_via_tag` | No | [send] Deprecated; accepted for compatibility but ignored. API sends always
|
|
56
|
+
| `--skip_sent_via_tag` | No | [send] Deprecated; accepted for compatibility but ignored. API sends always carry the sender provenance marker GenTeam shows next to the sender's name. (boolean) |
|
|
57
57
|
| `--message_id` | No | [react] comet_message_id of the message to react to (from genteam_read). \| [delete] comet_message_id of YOUR OWN message to delete (from genteam_read). \| [update] comet_message_id of YOUR OWN message to edit (from genteam_read). \| [mark_read] Optional comet_message_id (from genteam_read). Marking read: the highest message to mark read (inclusive); omit to mark the whole conversation read. Marking unread: the first message to show as unread (group chats and DMs only — threads take no anchor); omit to anchor at the latest message. \| [forward] comet_message_id of the message to forward (from genteam_read of the source conversation). \| [pins] pin/unpin: comet_message_id of the target message (from genteam_read or op=list output). (string) |
|
|
58
58
|
| `--emoji` | No | [react] The emoji as a unicode character itself, e.g. '👍' or '🎉', or a GenTeam custom emoji id exactly as it appears in a message's reactions (e.g. ':ack-x7k2:'). Slack shortcodes like ':thumbsup:' are not accepted. (string) |
|
|
59
59
|
| `--remove` | No | [react] true removes YOUR OWN existing reaction instead of adding one. Default: false. (boolean, default: `False`) |
|
|
@@ -46,7 +46,7 @@ gsk google_calendar [options]
|
|
|
46
46
|
| `--description` | No | [create] Description or details of the event. Supports restricted HTML formatting (safe subset rendered via Vue v-html): <a>, <b>, <strong>, <i>, <em>, <u>, <br>, <p>, <ul>, <ol>, <li>, <span>, <img>. Use for links, bold, italics, lists, line breaks, images. \| [modify] New event description (optional). Supports the same restricted HTML subset as create. (string) |
|
|
47
47
|
| `--time_zone` | No | [create] Time zone for the event (e.g., 'GMT-07:00') \| [modify] Optional time zone for start/end (e.g. 'America/Los_Angeles' or 'GMT-07:00'). (string) |
|
|
48
48
|
| `--time_zone_name` | No | [create] Time zone name for the event (e.g., 'America/Los_Angeles') (string) |
|
|
49
|
-
| `--attendees` | No | [create] List of email addresses of attendees \| [modify]
|
|
49
|
+
| `--attendees` | No | [create] List of email addresses of attendees \| [modify] Legacy compatibility field: must include every current attendee; omissions are rejected. Use add_attendees / remove_attendees for explicit membership changes. Mutually exclusive with those delta parameters. (array) |
|
|
50
50
|
| `--calendar_id` | No | [create] The calendar identifier. Use 'primary' for the user's primary calendar or provide a specific calendar email address (string) |
|
|
51
51
|
| `--event_id` | No | [create] The event identifier. If the request is to modify an existing event, provide the event_id of the event to be modified. If the request is to create a new event, leave this field empty. \| [respond] Google Calendar event id of the invitation to respond to. For a recurring-series instance, the id has shape '<seriesId>_<startTs>'. \| [modify] Google Calendar event id to modify. For a recurring-series instance, the id has shape '<seriesId>_<startTs>'. \| [delete] The Google Calendar event ID to delete (string) |
|
|
52
52
|
| `--recurrence` | No | [create] Recurrence rules (e.g., ['RRULE:FREQ=DAILY;COUNT=2']) \| [modify] New recurrence rules (e.g. ['RRULE:FREQ=WEEKLY;BYDAY=MO']). Only honored when scope=series. (array) |
|
|
@@ -47,7 +47,7 @@ gsk outlook_calendar [options]
|
|
|
47
47
|
| `--description` | No | [create] Description or details of the event. Supports restricted HTML formatting (safe subset rendered via Vue v-html): <a>, <b>, <strong>, <i>, <em>, <u>, <br>, <p>, <ul>, <ol>, <li>, <span>, <img>. Use for links, bold, italics, lists, line breaks, images. \| [modify] New event body (HTML allowed — Outlook renders the same restricted subset as create). (string) |
|
|
48
48
|
| `--time_zone` | No | [create] Time zone for the event (e.g., 'GMT-07:00') \| [respond] Optional time zone used for the proposed_new_time payload (IANA name, e.g. 'America/Los_Angeles'). Defaults to UTC. \| [modify] Optional IANA time zone for start/end (e.g. 'America/Los_Angeles'). Defaults to UTC. (string) |
|
|
49
49
|
| `--time_zone_name` | No | [create] Time zone name for the event (e.g., 'America/Los_Angeles') (string) |
|
|
50
|
-
| `--attendees` | No | [create] List of email addresses of attendees \| [modify]
|
|
50
|
+
| `--attendees` | No | [create] List of email addresses of attendees \| [modify] Legacy compatibility field: must include every current attendee; omissions are rejected. Use add_attendees / remove_attendees for explicit membership changes. Mutually exclusive with those delta parameters. (array) |
|
|
51
51
|
| `--calendar_id` | No | [create] The calendar identifier. Use 'primary' for the user's primary calendar or provide a specific calendar email address (string) |
|
|
52
52
|
| `--event_id` | No | [create] The event identifier. If the request is to modify an existing event, provide the event_id of the event to be modified. If the request is to create a new event, leave this field empty. \| [respond] Outlook event id of the invitation to respond to (from a prior outlook_calendar_list call). \| [modify] Outlook event id to modify. \| [delete] The Outlook Calendar event ID to delete (string) |
|
|
53
53
|
| `--recurrence` | No | [create] Recurrence pattern for the event (object) |
|
|
@@ -67,7 +67,7 @@ Drive has two containers and one view:
|
|
|
67
67
|
Things you must know about Drive:
|
|
68
68
|
|
|
69
69
|
- **Address = drive + entry.** Every file and folder is identified by the drive it lives in plus its entry id. Drive URLs carry both (`/s/<drive>/<entry>`, `?id=<entry>&link=<drive>`). `gsk drive ls --url <pasted link>` or `--id <entry> --owner <drive>` resolves them for you.
|
|
70
|
-
- **Shortcuts.** A share card, or a linked entry inside a Team Drive, is a pointer to a source entry in another drive. It has no bytes of its own. Resolve it first — a `LinkNode` row returns `link_workspace_id`; reuse it as `--source` — and read the source. Permission comes from the source, never from the pointer.
|
|
70
|
+
- **Shortcuts.** A share card, or a linked entry inside a Team Drive, is a pointer to a source entry in another drive. It has no bytes of its own. Resolve it first — a `LinkNode` row returns `link_workspace_id`; reuse it as `--source` — and read the source. Permission comes from the source, never from the pointer. A `web` link (`link_target_kind: web`) points at an external page instead: it has no bytes and no source to resolve, so read the page with `gsk crawl <link_target_url>` (the SSRF-guarded crawler); never fetch it with curl or wget.
|
|
71
71
|
- **Reusable addresses.** `ls` and `find` return `entry_locator`: reuse its workspace/source/path together. Reading a shortcut by that path follows the live source. Its `content_source_id` or `link_workspace_id` instead mounts the source at `/`; do not append the alias's old path to that mount. `rm` and `move` on the alias operate on the shortcut itself. A child path operates inside the source, with fresh authorization.
|
|
72
72
|
- **Only a name is known.** Use `gsk drive find --scope all -q <name>` across My Drive and shared sources; `mine` or `shared` narrows it. A matching name is a candidate, not a unique identity: compare source and parent path. Inspect `coverage`: shared searches have source/result limits and indexed content may be incomplete. Empty results do not prove absence. A source-specific folder search covers its immediate children.
|
|
73
73
|
- **GenTeam channel and DM drives are flat.** No `mkdir`, no directory upload, no moving native entries into subfolders (`GENTEAM_DRIVE_FLAT_LAYOUT`). Members with content-edit access upload at the root; moving, renaming or deleting native files needs conversation admin.
|
|
@@ -71,6 +71,28 @@ gsk img "Enhance this" -i ./photo.png -o ./result.png
|
|
|
71
71
|
|
|
72
72
|
Use `-o` / `--output-file` to save generated results directly to a local file.
|
|
73
73
|
|
|
74
|
+
### Task attachments (`gsk task create ... --file`)
|
|
75
|
+
|
|
76
|
+
`--file <paths|urls...>` is a CLI-only option (the server's `file_urls` parameter is filled from it and is not listed in the gsk-create-task skill). Local paths are uploaded first; `http(s)://`, `aidrive://` and file-wrapper URLs pass through unchanged. The files are NOT pasted into the query: each one becomes a chat attachment on the task's first user message (same shape as attaching a file in the web composer), so images are vision-visible and other files arrive as named documents the agent reads with its tools.
|
|
77
|
+
|
|
78
|
+
```bash
|
|
79
|
+
# Media agent: reference images, video, audio and documents all travel as attachments
|
|
80
|
+
gsk task create media_agent --task_name "Solar teaser" \
|
|
81
|
+
--query "One key-art image in the style of the reference image, then a 5-second clip from it with the attached voice line as the audio track" \
|
|
82
|
+
--instructions "Audience: kids 8-12; follow brief.pdf" \
|
|
83
|
+
--file ./ref.png ./clip.mp4 ./voice.mp3 ./brief.pdf
|
|
84
|
+
|
|
85
|
+
# Sheets: CSV / Excel files are imported into the workbook
|
|
86
|
+
gsk task create sheets --task_name "Q3" --query "Summarize the data" --instructions "One chart" --file ./sales.csv
|
|
87
|
+
```
|
|
88
|
+
|
|
89
|
+
| Task Type | What `--file` means |
|
|
90
|
+
|-----------|---------------------|
|
|
91
|
+
| media_agent | Reference material (images / video / audio / documents) the Media agent can view, transcribe, read and reuse |
|
|
92
|
+
| design | Structured context attachments (use `--image` for ordered visual references) |
|
|
93
|
+
| sheets | CSV / Excel import |
|
|
94
|
+
| other types | Attached as documents; mention them in the query so the agent knows what to do with them |
|
|
95
|
+
|
|
74
96
|
### Task agent export (`gsk task create ... -o`)
|
|
75
97
|
|
|
76
98
|
For task agents (docs, slides, sheets), `-o` follows the run live (implies `--follow`) and exports to standard file formats once it completes:
|
|
@@ -33,7 +33,7 @@ gsk video [options]
|
|
|
33
33
|
| `<query>` (positional) | No | Detailed, self-contained description of the video to generate, written in English. This text is sent to the video model verbatim as the generation prompt, so put all the visual detail here — but NEVER output settings: text like '16:9' or '4 seconds' in the prompt may be rendered literally into the scene. Pass aspect ratio via the `aspect_ratio` parameter and length via `duration`. duration is recommended to be 5 ~ 10 seconds. For image-to-image transitions (user provides start and end frames) e.g., 'A timelapse of a flower blooming in a garden'. Omit when passing the prompt by `query_file` instead. Or omit on a draft_task_id finalize (it reuses the draft's prompt). (string) |
|
|
34
34
|
| `--query_file` | No | Optional. Repo-relative path to a UTF-8 text file containing the prompt, used verbatim in place of `query` (lets you pass a long prompt by path instead of inlining it). Requires `repo_id`. When set, `query` may be omitted. (string) |
|
|
35
35
|
| `--file_name` | No | The name of the video to generate. (string, null) |
|
|
36
|
-
| `-m`, `--model` | No | The model to use for video generation (required for a render; omit only with task_id). kling/v3: Latest Kling V3. Supports native audio, but generates a SILENT video unless audio_enable=true. image_urls are frames, not references: 1 image = first frame (i2v); 2 images = first + last frame (the 2nd is the END frame). No reference mode (use kling/o3 with reference_mode for reference-driven generation). Pro/Standard/Turbo tiers (set via the tier parameter; turbo is the fastest arm but has no audio and no first/last-frame control). Supported ratios: 16:9, 9:16, 1:1. Duration: 3-15s. kling/o3: Kling O3. REQUIRES at least one input image or a video_url (no text-to-video). image_urls are frames (1 = first, 2 = first + last) unless reference_mode=true: style/character references, max 4 total including elements. video_url switches to video-to-video (video_mode 'reference' = motion guide, 'edit' = restyle in footage). Native audio via audio_enable. Pro/Standard tiers. Supported ratios: 16:9, 9:16, 1:1. Duration: 3-15s. gemini/veo3.1: Gemini Veo 3.1 - Latest version with enhanced quality and features. Supports text-to-video and image-to-video generation with improved quality. image_urls are frames unless reference_mode: 1 image = first frame, exactly 2 = first/last frame transition, 3 (or reference_mode=true) = style/character references (max 3, 8s only). video_url (MP4 source) = video EXTENSION mode: continues from the last frame, 7s, up to 1080p. Duration: 4s/6s/8s. Supported ratios: 16:9, 9:16. Supports tier (standard \| fast). Resolution: 720p, 1080p, 4k. minimax/h3: MiniMax H3 2K video model. Supports text-to-video, image-to-video with first/last frame control, and reference-to-video (up to 9 reference images, 3 reference videos, 3 reference audio clips — refer to them in the prompt as 'Image 1', 'Video 1', 'Audio 1', not @-style). Keyframes and references are mutually exclusive. Supported ratios: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-15s. Resolution: 768p or 2K (default). minimax/h3-max: MiniMax H3 Max, a distinct post-trained model with faster generation, stronger prompt adherence, and improved aesthetics. Supports text, first/middle/last frame keyframes, and image/video/audio references — keyframes and references COMBINE on this model (set keyframes via first_frame_url/last_frame_url/middle_image_url + middle_frame_time; the middle frame needs an explicit first_frame_url AND last_frame_url — image_urls are all references here — native 480p/768p, and at least one reference). tier=turbo selects the faster, half-price text/image arm; reference and middle-frame inputs use standard. Duration: 5-15s. Resolution: 480p or 768p. Reference requests also incur a reference-token fee on top of the output charge — a reference video especially can cost more than the output itself. wan/v2.7: Wan v2.7 with enhanced motion smoothness and scene fidelity. Supports text-to-video, image-to-video, reference-to-video, and edit-video. Ratios: 16:9, 9:16, 1:1. Duration: 5s. Resolution: 480p, 720p vidu/q4: Vidu Q4, Vidu's next-generation flagship with native audio. REQUIRES an image (no text-to-video): 1 image_urls entry = first frame (image-to-video, framing follows the image); reference_mode (or 2+ images, or audio_urls) = reference-to-video with up to 12 reference images and 3 voice clips, addressed in the prompt as [@reference_image_1].. / [reference_audio_1].. No end frame, no tiers. Supported ratios (reference mode): 16:9, 9:16, 4:3, 3:4, 1:1. Duration: 3-16s. Resolution: 540p, 720p, 1080p, 2K, 4Krunway/gen4_turbo: A model for generate video with high quality, fast.Supported ratios: 5:3, 3:5. Only support i2v. Duration: 5s, 10spixverse/v6: PixVerse V6 latest model with lifelike motion, richer skin detail, real emotions. Full cinematic control including choreography and camera. Supports text-to-video, image-to-video, transition, and video extend generation. VFX, time-lapse, transformation scenes, product demos, 360° views, multi-shot storytelling. Extend: pass video_url to extend an existing video. Supported ratios: 16:9, 9:16, 4:3, 1:1, 3:4. Duration: 5s, 8s. pixverse/c1: PixVerse C1 with native audio (audio_enable=true), text-to-video, image-to-video, transitions (2 images = first/last frame) and reference-to-video (reference_mode with 1-7 image_urls; address them in the prompt as @image1..@image7). No video extend. Supported ratios: 16:9, 9:16, 4:3, 3:4, 1:1, 3:2, 2:3, 21:9. Duration: 1-15s. Resolution: 360p, 540p, 720p, 1080p. fal-ai/bytedance/seedance-2.0: Bytedance Seedance 2.0 model for highest quality video generation with native audio and lip-sync. Supports text-to-video and image-to-video with first/last frame control. Supported ratios: 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 4-15s. Resolution: 480p, 720p, or 1080p on the standard tier (~2.25x cost; fast/mini cap at 720p). tier picks the variant: standard, fast, or mini (cheapest, ~half the standard price). Also supports reference-to-video with up to 9 images, 3 videos, or 3 audio refs via settings.fal-ai/topaz/upscale/video: Topaz Labs Video Upscaler, the preferred upscaler: 2x enhancement of an existing video (source up to 60s) with real detail recovery. Requires video_url parameter. Does NOT generate new videos. Use when user wants to improve quality of existing videos.fal-ai/bytedance-upscaler/upscale/video: ByteDance Video Upscaler for enhancing video quality. Upscales videos to higher resolutions (2k) Requires video_url parameter. Does NOT generate new videos. Use when user wants to improve quality of existing videos.xai/grok-imagine-video: xAI Grok Imagine Video v1.5 for high-quality video generation. Supports text-to-video and image-to-video generation. 2+ image_urls (or reference_mode=true) = style/character references, up to 7 (reference mode caps at 720p). video_url (MP4, 2-15s source) = video EXTENSION mode: continues from the last frame, 2-10s extension, 720p. 480p/720p/1080p output. Supported ratios: 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16. Duration: 1-15s. tier=lite runs the lightweight low-cost 1.5 Lite arm (text-to-video and image-to-video only — no references, no extension). alibaba/happy-horse: Alibaba Happy Horse video model. High-quality generation from text, or image-to-video from a single first-frame image (one image_url without reference_mode). Supported ratios: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21, 5:4, 4:5. Duration: 3-15s. Resolution: 720p, 1080p.alibaba/happy-horse/reference-to-video: Alibaba Happy Horse reference-to-video. Generate video from 1-9 reference images; address subjects in the prompt as character1..character9 (order matches image_urls). Requires image_urls. Supported ratios: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21, 5:4, 4:5. Duration: 3-15s. Resolution: 720p, 1080p.alibaba/happy-horse/video-edit: Alibaba Happy Horse video-edit. Edit a source video using a text prompt; optional 1-5 reference images addressed as @Image1..@Image5. Requires video_url (MP4/MOV, 3-60s). Output preserves source aspect ratio, capped at 15s. Resolution: 720p, 1080p.gemini/omni-flash: Google's multimodal video model with native audio. Supports text-to-video, image-to-video (1 image as first frame), reference-to-video (set reference_mode with image_urls to use images as subject/style references — even a single image; 2+ images are references by default), and video editing (pass video_url to edit an existing clip — restyle, restage, add objects). Bind image roles in the prompt with <FIRST_FRAME> / <IMAGE_REF_0>. Ratios: 16:9, 9:16. Duration: 3-10s. Resolution: 720p. bfl/flux-3-preview-high: BFL FLUX 3 video model with native audio. Supports text-to-video and 1-10 ordered keyframes, or continuation from one source video selected with input_mode=start_video. Ratios: auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-20s. Resolution: 720p, 1080p, 2K, 4K (2K/4K not available when continuing a video). Only one input family may be used per request.bfl/flux-3-max: FLUX 3 Max, fal's post-trained FLUX 3 video model: top-ranked image-to-video quality, renders a 10s clip in about 12s, native audio. Text-to-video and 1-10 ordered keyframes (2 images = first and last frame). No video continuation, no draft, no auto duration. Ratios: auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-20s (explicit). Resolution: 720p, 1080p.fal-ai/bytedance/seedance-2.5: Bytedance Seedance 2.5 for native 30-second single-shot video with synchronized audio. Supports text-to-video and image-to-video with first/last frame control. No tier variants. Supported ratios: auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 4-30s, or 0 = auto (needs video refs). Resolution: 480p, 720p, or 1080p (~2.46x cost). Also supports reference-to-video (up to 30 images, 10 videos, 10 audio refs; no image required), video EDITING and EXTENSION — source in video_urls, the prompt describes the change or continuation.wan/v3.0: Alibaba Wan 3.0 — next-gen open-weight Wan with native audio and single-pass clips up to 30s. Supports text-to-video, image-to-video with first/last frame control, and reference-to-video (reference_mode with up to 10 reference images, 5 videos and 5 audio clips, each ref modality <=15s combined — address them positionally in the prompt as 'Image 1', 'Video 1', 'Audio 1'). tier=fast runs the accelerated Prime arm (~1.4x cost). Supported ratios: adaptive, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 2-30s. Resolution: 480p, 720p, 1080p. Duration ranges above are hard caps; out-of-range values are clamped to the nearest supported value. (string, one of: kling/v3, kling/o3, gemini/veo3.1, minimax/h3, minimax/h3-max, wan/v2.7, vidu/q4, runway/gen4_turbo, pixverse/v6, pixverse/c1, fal-ai/bytedance/seedance-2.0, fal-ai/topaz/upscale/video, fal-ai/bytedance-upscaler/upscale/video, xai/grok-imagine-video, alibaba/happy-horse, alibaba/happy-horse/reference-to-video, alibaba/happy-horse/video-edit, gemini/omni-flash, bfl/flux-3-preview-high, bfl/flux-3-max, fal-ai/bytedance/seedance-2.5, wan/v3.0) |
|
|
36
|
+
| `-m`, `--model` | No | The model to use for video generation (required for a render; omit only with task_id). kling/v3: Latest Kling V3. Supports native audio, but generates a SILENT video unless audio_enable=true. image_urls are frames, not references: 1 image = first frame (i2v); 2 images = first + last frame (the 2nd is the END frame). No reference mode (use kling/o3 with reference_mode for reference-driven generation). Pro/Standard/Turbo tiers (set via the tier parameter; turbo is the fastest arm but has no audio and no first/last-frame control). Supported ratios: 16:9, 9:16, 1:1. Duration: 3-15s. kling/o3: Kling O3. REQUIRES at least one input image or a video_url (no text-to-video). image_urls are frames (1 = first, 2 = first + last) unless reference_mode=true: style/character references, max 4 total including elements. video_url switches to video-to-video (video_mode 'reference' = motion guide, 'edit' = restyle in footage). Native audio via audio_enable. Pro/Standard tiers. Supported ratios: 16:9, 9:16, 1:1. Duration: 3-15s. gemini/veo3.1: Gemini Veo 3.1 - Latest version with enhanced quality and features. Supports text-to-video and image-to-video generation with improved quality. image_urls are frames unless reference_mode: 1 image = first frame, exactly 2 = first/last frame transition, 3 (or reference_mode=true) = style/character references (max 3, 8s only). video_url (MP4 source) = video EXTENSION mode: continues from the last frame, 7s, up to 1080p. Duration: 4s/6s/8s. Supported ratios: 16:9, 9:16. Supports tier (standard \| fast). Resolution: 720p, 1080p, 4k. minimax/h3: MiniMax H3 2K video model. Supports text-to-video, image-to-video with first/last frame control, and reference-to-video (up to 9 reference images, 3 reference videos, 3 reference audio clips — refer to them in the prompt as 'Image 1', 'Video 1', 'Audio 1', not @-style). Keyframes and references are mutually exclusive. Supported ratios: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-15s. Resolution: 768p or 2K (default). minimax/h3-max: MiniMax H3 Max, a distinct post-trained model with faster generation, stronger prompt adherence, and improved aesthetics. Supports text, first/middle/last frame keyframes, and image/video/audio references — keyframes and references COMBINE on this model (set keyframes via first_frame_url/last_frame_url/middle_image_url + middle_frame_time; the middle frame needs an explicit first_frame_url AND last_frame_url — image_urls are all references here — native 480p/768p, and at least one reference). tier=turbo selects the faster, half-price text/image arm; reference and middle-frame inputs use standard. Duration: 5-15s. Resolution: 480p or 768p. Reference requests also incur a reference-token fee on top of the output charge — a reference video especially can cost more than the output itself. wan/v2.7: Wan v2.7 with enhanced motion smoothness and scene fidelity. Supports text-to-video, image-to-video, reference-to-video, and edit-video. Ratios: 16:9, 9:16, 1:1. Duration: 5s. Resolution: 480p, 720p vidu/q4: Vidu Q4, Vidu's next-generation flagship with native audio. REQUIRES an image (no text-to-video): 1 image_urls entry = first frame (image-to-video, framing follows the image); reference_mode (or 2+ images, or audio_urls) = reference-to-video with up to 12 reference images and 3 voice clips, addressed in the prompt as [@reference_image_1].. / [reference_audio_1].. No end frame, no tiers. Supported ratios (reference mode): 16:9, 9:16, 4:3, 3:4, 1:1. Duration: 3-16s. Resolution: 540p, 720p, 1080p, 2K, 4Krunway/gen4_turbo: A model for generate video with high quality, fast.Supported ratios: 5:3, 3:5. Only support i2v. Duration: 5s, 10spixverse/v6: PixVerse V6 latest model with lifelike motion, richer skin detail, real emotions. Full cinematic control including choreography and camera. Supports text-to-video, image-to-video, transition, and video extend generation. VFX, time-lapse, transformation scenes, product demos, 360° views, multi-shot storytelling. Extend: pass video_url to extend an existing video. Supported ratios: 16:9, 9:16, 4:3, 1:1, 3:4. Duration: 5s, 8s. pixverse/c1: PixVerse C1 with native audio (audio_enable=true), text-to-video, image-to-video, transitions (2 images = first/last frame) and reference-to-video (reference_mode with 1-7 image_urls; address them in the prompt as @image1..@image7). No video extend. Supported ratios: 16:9, 9:16, 4:3, 3:4, 1:1, 3:2, 2:3, 21:9. Duration: 1-15s. Resolution: 360p, 540p, 720p, 1080p. fal-ai/bytedance/seedance-2.0: Bytedance Seedance 2.0 model for highest quality video generation with native audio and lip-sync. Supports text-to-video and image-to-video with first/last frame control. Supported ratios: 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 4-15s. Resolution: 480p, 720p, or 1080p on the standard tier (~2.25x cost; fast/mini cap at 720p). tier picks the variant: standard, fast, or mini (cheapest, ~half the standard price). Also supports reference-to-video with up to 9 images, 3 videos, or 3 audio refs via settings.fal-ai/topaz/upscale/video: Topaz Labs Video Upscaler, the preferred upscaler: 2x enhancement of an existing video (source up to 60s) with real detail recovery. Requires video_url parameter. Does NOT generate new videos. Use when user wants to improve quality of existing videos.fal-ai/bytedance-upscaler/upscale/video: ByteDance Video Upscaler for enhancing video quality. Upscales videos to higher resolutions (2k) Requires video_url parameter. Does NOT generate new videos. Use when user wants to improve quality of existing videos.xai/grok-imagine-video: xAI Grok Imagine Video v1.5 for high-quality video generation. Supports text-to-video and image-to-video generation. 2+ image_urls (or reference_mode=true) = style/character references, up to 7 (reference mode caps at 720p). video_url (MP4, 2-15s source) = video EXTENSION mode: continues from the last frame, 2-10s extension, 720p. 480p/720p/1080p output. Supported ratios: 16:9, 4:3, 3:2, 1:1, 2:3, 3:4, 9:16. Duration: 1-15s. tier=lite runs the lightweight low-cost 1.5 Lite arm (text-to-video and image-to-video only — no references, no extension). alibaba/happy-horse: Alibaba Happy Horse video model. High-quality generation from text, or image-to-video from a single first-frame image (one image_url without reference_mode). Supported ratios: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21, 5:4, 4:5. Duration: 3-15s. Resolution: 720p, 1080p.alibaba/happy-horse/reference-to-video: Alibaba Happy Horse reference-to-video. Generate video from 1-9 reference images; address subjects in the prompt as character1..character9 (order matches image_urls). Requires image_urls. Supported ratios: 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, 9:21, 5:4, 4:5. Duration: 3-15s. Resolution: 720p, 1080p.alibaba/happy-horse/video-edit: Alibaba Happy Horse video-edit. Edit a source video using a text prompt; optional 1-5 reference images addressed as @Image1..@Image5. Requires video_url (MP4/MOV, 3-60s). Output preserves source aspect ratio, capped at 15s. Resolution: 720p, 1080p.gemini/omni-flash: Google's multimodal video model with native audio. Supports text-to-video, image-to-video (1 image as first frame), reference-to-video (set reference_mode with image_urls to use images as subject/style references — even a single image; 2+ images are references by default), and video editing (pass video_url to edit an existing clip — restyle, restage, add objects). Bind image roles in the prompt with <FIRST_FRAME> / <IMAGE_REF_0>. Ratios: 16:9, 9:16. Duration: 3-10s. Resolution: 720p. kling/v4: Kling 4.0 — fast 720p/1080p generation. Text-to-video, image-to-video (1 image = start frame, 2 images = start + end frame), reference-to-video (reference_mode or 3+ images: up to 10 image_urls as @Image1.., up to 5 video_urls as @Video1.., up to 7 elements as @Element1..) and video-to-video editing (pass video_url — the prompt is the edit instruction, @Video1 addresses the source clip). No extension mode. Supported ratios: 16:9, 9:16, 1:1, 21:9. Duration: 3-30s. Resolution: 720p, 1080p. bfl/flux-3-preview-high: BFL FLUX 3 video model with native audio. Supports text-to-video and 1-10 ordered keyframes, or continuation from one source video selected with input_mode=start_video. Ratios: auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-20s. Resolution: 720p, 1080p, 2K, 4K (2K/4K not available when continuing a video). Only one input family may be used per request.bfl/flux-3-max: FLUX 3 Max, fal's post-trained FLUX 3 video model: top-ranked image-to-video quality, renders a 10s clip in about 12s, native audio. Text-to-video and 1-10 ordered keyframes (2 images = first and last frame). No video continuation, no draft, no auto duration. Ratios: auto, 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 5-20s (explicit). Resolution: 720p, 1080p.fal-ai/bytedance/seedance-2.5: Bytedance Seedance 2.5 for native 30-second single-shot video with synchronized audio. Supports text-to-video and image-to-video with first/last frame control. No tier variants. Supported ratios: auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 4-30s, or 0 = auto (needs video refs). Resolution: 480p, 720p, or 1080p (~2.46x cost). Also supports reference-to-video (up to 30 images, 10 videos, 10 audio refs; no image required), video EDITING and EXTENSION — source in video_urls, the prompt describes the change or continuation.wan/v3.0: Alibaba Wan 3.0 — next-gen open-weight Wan with native audio and single-pass clips up to 30s. Supports text-to-video, image-to-video with first/last frame control, and reference-to-video (reference_mode with up to 10 reference images, 5 videos and 5 audio clips, each ref modality <=15s combined — address them positionally in the prompt as 'Image 1', 'Video 1', 'Audio 1'). tier=fast runs the accelerated Prime arm (~1.4x cost). Supported ratios: adaptive, 16:9, 4:3, 1:1, 3:4, 9:16. Duration: 2-30s. Resolution: 480p, 720p, 1080p. Duration ranges above are hard caps; out-of-range values are clamped to the nearest supported value. (string, one of: kling/v3, kling/o3, gemini/veo3.1, minimax/h3, minimax/h3-max, wan/v2.7, vidu/q4, runway/gen4_turbo, pixverse/v6, pixverse/c1, fal-ai/bytedance/seedance-2.0, fal-ai/topaz/upscale/video, fal-ai/bytedance-upscaler/upscale/video, xai/grok-imagine-video, alibaba/happy-horse, alibaba/happy-horse/reference-to-video, alibaba/happy-horse/video-edit, gemini/omni-flash, kling/v4, bfl/flux-3-preview-high, bfl/flux-3-max, fal-ai/bytedance/seedance-2.5, wan/v3.0) |
|
|
37
37
|
| `-i`, `--image_urls` | No | The URLs of the images to use as reference key frames for the video generation. For single image models, provide 1 image. For multi-image models (like kling/v1.6/pro/elements), provide 1-4 images in sequence order. For start-end models (like vidu, pixverse), provide exactly 2 images (start and end frames). (default is [], if the task is based on one or more reference images, it is required) IMPORTANT: without reference_mode, the first image becomes the video's literal opening frame — identity markers visible in it (flags, jerseys, text, logos, faces) persist in the output and the text prompt CANNOT override them. When generating a batch of subject-specific clips (per-country / per-person / per-product), each clip needs an image matching its own subject — never reuse another subject's image. (array) |
|
|
38
38
|
| `--first_frame_url` | No | MiniMax H3 Max keyframe control: image URL the video opens on exactly, ALONGSIDE reference_mode references (the output canvas follows this image). Without reference_mode the first image_urls entry already is the first frame; in reference_mode (including any middle-frame request) it is REQUIRED, because image_urls entries are all subject references. H3 Max only: other models ignore it, and base minimax/h3 rejects it (frames there come from image_urls). (string) |
|
|
39
39
|
| `--middle_image_url` | No | MiniMax H3 Max keyframe control: image the video passes through at middle_frame_time. Requires an explicit first frame (first_frame_url) AND last frame (last_frame_url) — a middle frame forces reference mode, where every image_urls entry is a subject reference — plus native 480p/768p and at least one reference image/video/audio. H3 Max only: other models ignore it, and base minimax/h3 rejects it. (string) |
|