Skills

Skill 24 of 34

Generate Image

Generate or edit images with RunComfy AI image models including FLUX, Seedream and Nano Banana.

2 minutes · 458 words · 4 sections

Install

$ npx skills add runcomfy-com/skills --skill generate-image

$ npx skills add runcomfy-com/skills

» /plugin marketplace add runcomfy-com/skills

The first command installs just this skill, by the name in its SKILL.md; the second installs the whole repository.

Generate and edit images with RunComfy

Use the connected RunComfy MCP tools. The client may prefix tool names with the plugin/server name. If disconnected, direct the user to the client’s RunComfy authentication control. Never request an API token in chat or add one to a file.

Choose and review

Establish the subject, intended use, dimensions, number of images and any reference images. Reuse details already provided.
Use list_models with a targeted search and category such as text-to-image or image-to-image. Use list_model_categories if unsure of a category. Start with a small page and paginate when needed.
Match the user’s requested model family. Examples available through RunComfy include FLUX 2, Seedream 5.0 and Nano Banana 2. Use the actual returned model_id; do not construct version strings or assume a similarly named endpoint supports edits.
Call get_model for the selected ID. Read current input schema, required inputs, enums, defaults, pricing unit and output type. Do not reuse another model’s field names or hard-code remembered prices.
Check get_balance before paid work. Present the model, final prompt, reference inputs, dimensions/count and available cost estimate. Proceed within the user’s existing authorization and spending limit; ask only when scope or cost is missing or changed. Explain when pricing is an estimate.

Run and retrieve

For a file input, use an HTTPS URL accessible to the model service. A local path, browser-only blob: URL or authenticated gallery page is not a usable input URL. Do not publish a private image without authorization; if no supported upload route is available, explain what input is needed.
Call run_model once with model_id, schema-valid inputs and wait_for_completion: false. Record the returned request ID.
Poll get_model_request_status for that same ID with reasonable spacing. Retrieve get_model_request_result when finished. Use cancel_model_request only when the user requested cancellation or an agreed stop condition is reached.
Inspect the cancellation outcome: a running or finished job may be not_cancellable. A cancellation attempt is not proof the job stopped or that charges ended. Reconcile status/results and report this limitation; a monitoring deadline is not a guaranteed spending cap.
A timeout or interrupted response does not prove submission failed. Recover the existing request before considering a new paid call. Never silently submit a replacement or switch models.
Return the actual output URL and request ID. Display the image when supported; otherwise provide a direct link. Say whether pixels were inspected. A returned URL is not proof that the image rendered correctly.
Keep planned settings, returned settings and observed output properties separate. If the server fails or a result is unavailable, report the actual status and next recovery step.

Useful requests

“Compare current FLUX and Seedream image options for one square product image. Show inputs and price before running.”
“Edit this image to change only the background, using a supported image-editing model. Keep the product unchanged.”

AI image models (opens in a new tab) · Model API documentation (opens in a new tab) · MCP setup (opens in a new tab)

About this skill

Trigger

Generate or edit images with RunComfy AI image models including FLUX, Seedream and Nano Banana. Use for text-to-image, reference images, product photography, illustrations and image edits. Browse current models, review inputs and cost, then run and retrieve results.

The verbatim description from this skill’s front matter — the string an agent matches on to decide whether to load it.

Colophon for this issue

Colophon

How this issue was made

Source

github.com/runcomfy-com/skills, branch main, last pushed 15 September 2026.

Licence

MIT — the text of every skill is reproduced unmodified, frontmatter included, under the upstream licence.

Discovery

34 skills found by walking the repository tree for SKILL.md, not by matching a directory convention. 2 distinct layouts observed: */SKILL.md, plugins/runcomfy/skills/*/SKILL.md.

Issue colours

Resolved from a deterministic hash of the owner name. Two accent tones are generated per issue and each is proven against its own ground before it ships: a single accent that passes AA on both light and dark paper is arithmetically impossible.

Heading repairs

1 repair applied to this skill so the document has one h1 and no skipped levels:

Shifted “4 headings” from h1 to h2 so the skill title is the only h1.

Spec compliance

2 editorial notes across 2 of 34 skills. They are printed in the margin of each skill rather than as errors here.

AI Music — 1 note
Runcomfy CLI — 1 note

Images

Images inside a skill come from the upstream repository. Where the author gave no alternative text we mark the image decorative rather than inventing a description — a plausible caption we made up is worse than none for the reader who depends on it.

Marketplace

A plugin manifest is published at .claude-plugin/marketplace.json by RunComfy, declaring 1 plugin. It is read for editorial metadata only — never as the skill index, which is always the repository tree.

Signal

This repository is unranked on skills.sh, which says nothing about it: the leaderboard covers a fraction of published skills.

Agent surfaces

The whole issue is available as one markdown document at /runcomfy-com/skills.md, and each skill at its own .md URL.

Publication

Set by Skills Docs from the source repository. Body text is Literata at the reader’s chosen size and measure; code is Geist Mono. Nothing on this page was written by us except this paragraph.

Generate and edit images with RunComfy

Establish the subject, intended use, dimensions, number of images and any reference images. Reuse details already provided.

Use list_models with a targeted search and category such as text-to-image or image-to-image. Use list_model_categories if unsure of a category. Start with a small page and paginate when needed.

Match the user’s requested model family. Examples available through RunComfy include FLUX 2, Seedream 5.0 and Nano Banana 2. Use the actual returned model_id; do not construct version strings or assume a similarly named endpoint supports edits.

Call get_model for the selected ID. Read current input schema, required inputs, enums, defaults, pricing unit and output type. Do not reuse another model’s field names or hard-code remembered prices.

Check get_balance before paid work. Present the model, final prompt, reference inputs, dimensions/count and available cost estimate. Proceed within the user’s existing authorization and spending limit; ask only when scope or cost is missing or changed. Explain when pricing is an estimate.

For a file input, use an HTTPS URL accessible to the model service. A local path, browser-only blob: URL or authenticated gallery page is not a usable input URL. Do not publish a private image without authorization; if no supported upload route is available, explain what input is needed.

Call run_model once with model_id, schema-valid inputs and wait_for_completion: false. Record the returned request ID.

Poll get_model_request_status for that same ID with reasonable spacing. Retrieve get_model_request_result when finished. Use cancel_model_request only when the user requested cancellation or an agreed stop condition is reached.

Inspect the cancellation outcome: a running or finished job may be not_cancellable. A cancellation attempt is not proof the job stopped or that charges ended. Reconcile status/results and report this limitation; a monitoring deadline is not a guaranteed spending cap.

A timeout or interrupted response does not prove submission failed. Recover the existing request before considering a new paid call. Never silently submit a replacement or switch models.

Return the actual output URL and request ID. Display the image when supported; otherwise provide a direct link. Say whether pixels were inspected. A returned URL is not proof that the image rendered correctly.

Keep planned settings, returned settings and observed output properties separate. If the server fails or a result is unavailable, report the actual status and next recovery step.

“Compare current FLUX and Seedream image options for one square product image. Show inputs and price before running.”

“Edit this image to change only the background, using a supported image-editing model. Keep the product unchanged.”