Update to Dora AI Video
Snapshot Nov 1, 2026 · 12:10 UTC · version 1.0.1
Collection source: downloaded plugin package.
Instructions updated for get-started
Skill instructions — Now says: said what to make, ask only: "What do you want to create?" Offer no list of options. - If it is not clear whether they want a video or an image, ask only...
Observed in instructions or declared skills. Runtime behavior has not been tested.
Up to 10 plugins free · no card required.
All fields and original values.
Skill instructions
Oct 24, 2026 · 13:48 UTC
said, ask in one short message: vertical (9:16, for TikTok, Reels, Shorts), horizontal (16:9) or square (1:1); and for video, the length (4, 8 or 12 seconds) and whether they want sound. - Skip the questions when the user already gave th...
Nov 1, 2026 · 12:10 UTC
said what to make, ask only: "What do you want to create?" Offer no list of options. - If it is not clear whether they want a video or an image, ask only that. If it is clear, do not ask. - Then generate straight away with the tool defau...
Compare saved observations
Download comparison JSONSKILL.md text changes
− removed · + added
--- before +++ after @@ -9,17 +9,17 @@ ## Before generating -- If the user has not said, ask in one short message: vertical (9:16, for TikTok, Reels, Shorts), horizontal (16:9) or square (1:1); and for video, the length (4, 8 or 12 seconds) and whether they want sound. -- Skip the questions when the user already gave those details or asked for something quick. -- Use the defaults unless the user asks for more quality: `generate_image` with its default model at 1K, `generate_video` with its default model at 480p. +- If the user has not said what to make, ask only: "What do you want to create?" Offer no list of options. +- If it is not clear whether they want a video or an image, ask only that. If it is clear, do not ask. +- Then generate straight away with the tool defaults: vertical 9:16, 8 seconds, 480p, sound on when the model offers it; images at 1K. Change a setting only when the user asks. - If the user shares a photo of a person or a product, pass it as the reference image so the result keeps the same face or product. ## Generating 1. Call `generate_image` or `generate_video`. It returns a `job_id` right away. 2. Tell the user it is on its way: images take about a minute, videos about two. -3. Call `check_job` with the `job_id`, waiting about 30 seconds between checks, until the status is `done` or `failed`. -4. When it is done, give the user the `result_url`. If it failed, say so plainly and offer to try again; the coins for a failed generation are returned automatically. +3. Call `check_job` quietly with the `job_id`, waiting about 30 seconds between checks, until the status is `done` or `failed`. Do not end your turn before that. If you cannot keep waiting, say plainly that the result is not ready and to ask you for it in about two minutes. +4. When it is done, send one message with the link from `check_job`, as text. Do not embed the file, and repeat only the settings the tool returned. If it failed, say so plainly and offer to try again; the coins for a failed generation are returned automatically. ## When something stops the request
Full technical diff · 1 changed fields
changed /skill_md_contents
"---\nname: get-started\ndescription: Make a short video or image with Dora from a description or a photo, and deliver the result.\n---\n\n# Making videos and images with Dora\n\nDora turns a description or a photo into a short video or an image. Use it when the user wants to create, animate or restyle visual content.\n\n## Before generating\n\n- If the user has not said, ask in one short message: vertical (9:16, for TikTok, Reels, Shorts), horizontal (16:9) or square (1:1); and for video, the length (4, 8 or 12 seconds) and whether they want sound.\n- Skip the questions when the user already gave those details or asked for something quick.\n- Use the defaults unless the user asks for more quality: `generate_image` with its default model at 1K, `generate_video` with its default model at 480p.\n- If the user shares a photo of a person or a product, pass it as the reference image so the result keeps the same face or product.\n\n## Generating\n\n1. Call `generate_image` or `generate_video`. It returns a `job_id` right away.\n2. Tell the user it is on its way: images take about a minute, videos about two.\n3. Call `check_job` with the `job_id`, waiting about 30 seconds between checks, until the status is `done` or `failed`.\n4. When it is done, give the user the `result_url`. If it failed, say so plainly and offer to try again; the coins for a failed generation are returned automatically.\n\n## When something stops the request\n\n- `CONTENT_POLICY`: Dora does not create sexual content, content involving minors, graphic violence or self-harm. Say that this request cannot be made and offer a different direction. Do not try to rephrase the same request around the filter.\n- `INSUFFICIENT_CREDITS`: the user's Dora account does not have enough coins for this generation. Say so and share the link in `plans_url` for information. Do not encourage a purchase.\n- `AUTH_EXPIRED`: ask the user to reconnect Dora.\n\n## Series and characters\n\nFor a recurring character, ask for one or two clear photos of them and pass them as reference images on every generation, so the face and outfit stay consistent across clips.\n"
"---\nname: get-started\ndescription: Make a short video or image with Dora from a description or a photo, and deliver the result.\n---\n\n# Making videos and images with Dora\n\nDora turns a description or a photo into a short video or an image. Use it when the user wants to create, animate or restyle visual content.\n\n## Before generating\n\n- If the user has not said what to make, ask only: \"What do you want to create?\" Offer no list of options.\n- If it is not clear whether they want a video or an image, ask only that. If it is clear, do not ask.\n- Then generate straight away with the tool defaults: vertical 9:16, 8 seconds, 480p, sound on when the model offers it; images at 1K. Change a setting only when the user asks.\n- If the user shares a photo of a person or a product, pass it as the reference image so the result keeps the same face or product.\n\n## Generating\n\n1. Call `generate_image` or `generate_video`. It returns a `job_id` right away.\n2. Tell the user it is on its way: images take about a minute, videos about two.\n3. Call `check_job` quietly with the `job_id`, waiting about 30 seconds between checks, until the status is `done` or `failed`. Do not end your turn before that. If you cannot keep waiting, say plainly that the result is not ready and to ask you for it in about two minutes.\n4. When it is done, send one message with the link from `check_job`, as text. Do not embed the file, and repeat only the settings the tool returned. If it failed, say so plainly and offer to try again; the coins for a failed generation are returned automatically.\n\n## When something stops the request\n\n- `CONTENT_POLICY`: Dora does not create sexual content, content involving minors, graphic violence or self-harm. Say that this request cannot be made and offer a different direction. Do not try to rephrase the same request around the filter.\n- `INSUFFICIENT_CREDITS`: the user's Dora account does not have enough coins for this generation. Say so and share the link in `plans_url` for information. Do not encourage a purchase.\n- `AUTH_EXPIRED`: ask the user to reconnect Dora.\n\n## Series and characters\n\nFor a recurring character, ask for one or two clear photos of them and pass them as reference images on every generation, so the face and outfit stay consistent across clips.\n"
Full snapshot data
{
"description": "Make a short video or image with Dora from a description or a photo, and deliver the result.",
"included_files": [],
"name": "get-started",
"skill_md_contents": "---\nname: get-started\ndescription: Make a short video or image with Dora from a description or a photo, and deliver the result.\n---\n\n# Making videos and images with Dora\n\nDora turns a description or a photo into a short video or an image. Use it when the user wants to create, animate or restyle visual content.\n\n## Before generating\n\n- If the user has not said what to make, ask only: \"What do you want to create?\" Offer no list of options.\n- If it is not clear whether they want a video or an image, ask only that. If it is clear, do not ask.\n- Then generate straight away with the tool defaults: vertical 9:16, 8 seconds, 480p, sound on when the model offers it; images at 1K. Change a setting only when the user asks.\n- If the user shares a photo of a person or a product, pass it as the reference image so the result keeps the same face or product.\n\n## Generating\n\n1. Call `generate_image` or `generate_video`. It returns a `job_id` right away.\n2. Tell the user it is on its way: images take about a minute, videos about two.\n3. Call `check_job` quietly with the `job_id`, waiting about 30 seconds between checks, until the status is `done` or `failed`. Do not end your turn before that. If you cannot keep waiting, say plainly that the result is not ready and to ask you for it in about two minutes.\n4. When it is done, send one message with the link from `check_job`, as text. Do not embed the file, and repeat only the settings the tool returned. If it failed, say so plainly and offer to try again; the coins for a failed generation are returned automatically.\n\n## When something stops the request\n\n- `CONTENT_POLICY`: Dora does not create sexual content, content involving minors, graphic violence or self-harm. Say that this request cannot be made and offer a different direction. Do not try to rephrase the same request around the filter.\n- `INSUFFICIENT_CREDITS`: the user's Dora account does not have enough coins for this generation. Say so and share the link in `plans_url` for information. Do not encourage a purchase.\n- `AUTH_EXPIRED`: ask the user to reconnect Dora.\n\n## Series and characters\n\nFor a recurring character, ask for one or two clear photos of them and pass them as reference images on every generation, so the face and outfit stay consistent across clips.\n"
}SHA-256 of public snapshot: 8b39dadda8ef05b315f31fafe614db58dbc80fcc6295cd666c952b2bb1fd8b55