Veevid AI Video Generator

AI video generator — text to video, image to video, and reference-to-video via Veevid API. Supports Kling 3.0, Sora 2, Veo 3.1, LTX 2.3, Seedance, Wan 2.6, a...

Security Scan

The skill's requests and instructions are consistent with an AI video-generation integration that uses a local Veevid API key and accesses chat messages to locate images; nothing appears disproportionate or deceptive.

Purpose & Capability

The name/description (Veevid text/image/reference-to-video) align with the declared behavior: it expects a Veevid API key stored at ~/.config/veevid/api_key and calls veevid.ai endpoints. The only minor metadata mismatch: registry metadata shows 'Required config paths: [object Object]' which looks like a formatting error, but the SKILL.md clearly documents the API key path.

Instruction Scope

Runtime instructions are narrowly scoped to: (1) read a local API key file, (2) call veevid.ai endpoints (quote/generate/status/credits/storage), and (3) optionally read recent chat messages to find image attachments when the user requested image-to-video. Reading recent messages and posting quoted previews is justified for the I2V flow, but it is a privacy-sensitive action — the agent will access channel history and post previews back to the channel. The skill does not instruct reading other local files or sending data to third-party endpoints beyond veevid.ai.

Install Mechanism

Instruction-only skill with no install spec and no code to write to disk beyond the expected config directory (~/.config/veevid). This is low-risk from an installation perspective.

Credentials

The skill requests a single local config path (~/.config/veevid/api_key) for the Veevid API key and uses that key in Authorization headers. No unrelated environment variables, credentials, or broad system access are requested.

Persistence & Privilege

No 'always' privilege is requested; the skill runs only when invoked. It stores/reads its own config file under the user's home config directory, which is expected for an external API integration. It doesn't modify other skills or global agent settings.

Assessment

This skill appears to do exactly what it says: it uses a Veevid API key you store at ~/.config/veevid/api_key and calls veevid.ai endpoints to quote and generate videos. Before installing, consider: (1) the agent will read recent channel messages to locate images for image-to-video requests and will post quoted previews back to the channel — ensure you're comfortable with that privacy behavior in group chats; (2) keep your Veevid API key private (the skill reads it from a local file and uses it in requests); (3) there is no installer, so only the config file is created; and (4) the registry metadata contains a minor formatting error for the config path (harmless but worth noting). If you plan to use this skill in sensitive channels, limit where it can be invoked or remove agent permissions to read channel history.

All-in-one AI video generator: text to video, image to video, and reference-to-video via Veevid AI. One API, 10+ state-of-the-art video generation models including Kling 3.0, Sora 2, Veo 3.1, LTX 2.3, Seedance 1.5 Pro, and Wan 2.6.

Setup

1. Get your API Key at veevid.ai/settings/api-keys
2. Save the key locally:

mkdir -p ~/.config/veevid
echo "YOUR_API_KEY" > ~/.config/veevid/api_key

What You Can Say

These phrases trigger this skill:

• "Generate a video of a sunset over the ocean"
• "Turn this image into a video"
• "Make a 10-second product ad with Kling 3.0"
• "Create a vertical video for TikTok"
• "How much does it cost to generate a video?"
• "What AI video models are available?"
• "Check my credit balance"

Image Retrieval from Chat (Discord / Group Chats)

When the user requests image-to-video but no image URL is provided in the message:

1. Use message read (channel=discord, limit=10) to scan recent messages in the current channel.
2. Filter attachments by the sender's user id to avoid picking up someone else's image.
3. Collect all recent image attachments (content_type starts with image/) from the sender, ordered newest first.
4. Preview by quoting: Use message send with replyTo set to the message id of the image message. This quotes the original message with its image preview inline, so the user can visually confirm. Add text like "Is this the right image?" Don't just show the filename — filenames like IMG_20260203_xyz.jpg are meaningless.
5. Single image needed (standard I2V): quote the most recent image message and ask the user to confirm.
6. Multiple images needed (Kling 3.0 start+end frames, Veo 3.1 reference-to-video with 1-3 refs): quote each image message with a number label and ask the user which ones to use and in what order.
7. If not enough images are found, ask the user to send the missing ones or provide URLs.

This ensures image-to-video works seamlessly even when Discord doesn't forward attachments directly to the agent. Applies to all models, not just a specific one.

Important: Always Confirm Before Generating

Before calling the generate API, ALWAYS:

1. Call POST /api/quote with the billing-relevant params for your planned generation (see Step 1 below) — it returns the precise credit cost and current balance.
2. Tell the user which model you'll use and the exact cost from the quote response.
3. For image-to-video: also confirm the image is correct (show filename/URL).
4. Wait for the user to confirm before proceeding.

If the generate API returns 402, report the exact required and balance values from the response body to the user.

If balance is insufficient, suggest a cheaper model or link to veevid.ai/pricing.

Step 1 — Quote (get exact cost before generating)

/api/quote only needs the billing-relevant params: mode, generation_type, duration, video_quality, aspect_ratio, and model-specific options (e.g. model_version, generate_audio). It does not require prompt, image, or other content fields — those don't affect credit cost.

curl -X POST https://veevid.ai/api/quote \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "veo3",
    "generation_type": "text-to-video",
    "video_quality": "standard"
  }'

Returns: {"required_credits": 20, "current_balance": 451, "sufficient": true}

If sufficient is false, suggest a cheaper model before asking the user to confirm.

Step 2 — Generate Video (Text to Video)

curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A golden retriever running through sunflowers, cinematic lighting",
    "mode": "veo3",
    "aspect_ratio": "16:9",
    "video_quality": "standard"
  }'

Returns either:

• Async case: {"generation_id": "vg_xxx", "status": "processing", "video_url": null, "image_url": null}
• Sync case: {"generation_id": "vg_xxx", "status": "completed", "video_url": "https://..."}

If status is already completed and a video_url is present, return it to the user immediately instead of polling.

Generate Video — Image to Video

If the user provides an online image URL, pass it directly:

curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Gently animate this scene with camera movement",
    "mode": "veo3",
    "aspect_ratio": "16:9",
    "video_quality": "standard",
    "image": "https://example.com/photo.jpg",
    "generation_type": "image-to-video"
  }'

If the user provides a local file, the image URL isn't known yet — but credit cost doesn't depend on it, so quote first with billing params, then upload:

# Step 1: Quote (image URL not needed for pricing)
curl -X POST https://veevid.ai/api/quote \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{"mode": "veo3", "generation_type": "image-to-video", "video_quality": "standard"}'

# Step 2: Upload (only after user confirms)
curl -X POST https://veevid.ai/api/storage/upload \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -F "file=@/path/to/image.jpg"

# Returns: {"url": "https://cdn.veevid.ai/uploads/xxx.jpg", "key": "uploads/xxx.jpg"}

# Step 3: Generate with uploaded URL
curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Animate this scene",
    "mode": "veo3",
    "video_quality": "standard",
    "image": "https://cdn.veevid.ai/uploads/xxx.jpg",
    "generation_type": "image-to-video"
  }'

Step 3 — Poll Status

If the generate response is still processing, poll until status is completed:

curl -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  https://veevid.ai/api/video-generation/{generation_id}/status

Status values: processing → completed or failed. Typical time: 60-180 seconds.

Check Credit Balance

curl -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  https://veevid.ai/api/credits

Returns: {"success": true, "credits": 451, "balance": 451}

Available Models

Choose model based on the user's needs. Credit ranges depend on duration, resolution, and quality — use /api/quote for the exact cost before confirming with the user.

Model	mode	Strengths	Credit Range
Veo 3.1	veo3	Native audio, budget-friendly, fixed 8s	20 / 140 credits
Grok Imagine	grok-imagine	Native audio, fastest generation	10–60 credits
Veevid 1.0 Pro	veevid-1.0-pro	Audio, wide aspect ratios	12–288 credits
Seedance 1.5 Pro	seedance-1.5-pro	Audio, wide aspect ratios	12–288 credits
Kling 3.0	kling-3	Audio, multi-shot, 3-15s	48–495 credits
LTX 2.3	ltx-2-3	Audio, open-source, up to 4K (2160p)	48–960 credits
Kling 2.6	kling-2-6	Audio, cinematic, multi-shot	70–280 credits
Sora 2 Stable	sora2-stable	Native audio, best prompt accuracy, 4-20s	80–2000 credits
Sora 2	sora2	Native audio, 10-25s, storyboard/multi-scene	20–315 credits
Wan 2.6	wan-2-6	Audio, multi-shot, video-to-video	100–450 credits

Quick Model Picker

User wants...	Use this
Cheapest option	grok-imagine (10 cr at 480p/6s) or veo3 standard (20 cr)
Fastest result	grok-imagine or ltx-2-3 fast
Best quality	sora2-stable pro or kling-3
Longer clips (>15s)	sora2 (up to 25s)
Vertical/TikTok	ltx-2-3 (native portrait)
Product ads	kling-3 (multi-shot + text rendering)
YouTube content	veo3

Model Parameters

Each model has specific allowed values. See references/api-reference.md for the complete per-model parameter table.

Common parameters:

Parameter	Description
prompt	Scene description (max 5000 chars)
mode	Model selector (see table above)
duration	Video length in seconds (varies by model)
aspect_ratio	"16:9" | "9:16" | "1:1" (varies by model)
video_quality	Quality tier (varies by model — see api-reference.md)
generation_type	"text-to-video" | "image-to-video" | "reference-to-video" (Veo 3.1)
image	Image URL (required for image-to-video)
model_version	Sub-variant for some models (e.g. "kling-3-pro", "ltx-2.3-fast")

Error Codes

Code	Meaning	What to do
400	Invalid parameter	Check allowed values in api-reference.md
401	Invalid API Key	Verify key at veevid.ai/settings/api-keys
402	Insufficient credits	Show user the required vs balance from response, suggest cheaper model or veevid.ai/pricing
403	Account suspended / forbidden	Tell the user their account cannot access the API and they need support
404	Generation record not found	Check the generation_id and whether it belongs to this account
500	Server error	Wait a few seconds and retry

Example Conversations

Simple text-to-video:

User: "Make a 4-second video of a cat sleeping on a couch"
Agent: [calls /api/quote with mode=veo3, video_quality=standard]
Agent: "Note: Veo 3.1 always generates 8-second videos (duration is fixed). Cost: 20 credits. You have 311 credits. Proceed?"
User: "Yes"
Agent: [generates] "Done! Here's your 8-second video: [URL]."

Image-to-video:

User: "Turn this product photo into a short video" [attaches image]
Agent: [calls /api/quote with mode=veo3, generation_type=image-to-video]
Agent: "I'll upload your image and animate it with Veo 3.1. Cost: 20 credits. You have 200 credits. Proceed?"
User: "Go ahead"
Agent: [uploads image, generates] "Done! Here's your animated product video: [URL]"

Budget-conscious:

User: "Generate a video but keep it cheap"
Agent: [calls /api/quote with mode=grok-imagine, duration=6, video_quality=480p]
Agent: "The cheapest option is Grok Imagine at 10 credits for a 6-second 480p video. You have 50 credits. Want me to use that?"

Links

• Veevid AI
• Get API Key
• Pricing & Credits
• Blog & Tutorials