> For the complete documentation index, see [llms.txt](https://docs.flora.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.flora.ai/mcp/skills.md).

# FLORA MCP skills and plugins

The 57 FLORA MCP skills (18 workflow skills and 39 one-job skills), what each one does, and how to load them in Claude, Claude Code, ChatGPT, Codex or Cursor as plugins, raw files or on demand.

Skills are tested instructions that teach your agent a whole workflow on top of [FLORA MCP](https://flora.ai/mcp): batch generation, refine loops, product page sets, script to video and more. The server gives your agent the tools; a skill tells it which ones to call, in what order, and what goes wrong. Load one when the job is bigger than a single generation.

FLORA serves 57 skills through `flora_discover_skills`: the 18 workflow skills below and 39 [one-job skills](#one-job-skills) that each do one creative task. They are mirrored to the public [flora-mcp-skills](https://github.com/florafauna-ai/flora-mcp-skills) repository, one folder per skill, each with a `SKILL.md`. If a skill isn't in the repository yet, `flora_discover_skills` still serves it.

## The workflow skills

| Skill                           | What it does                                                                                                                                                                          |
| ------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `flora-start-here`              | Routes any FLORA request (image, video, audio, Technique, canvas, deck) to the right skill. Generates nothing itself.                                                                 |
| `flora-batch-generate`          | Turns a list (SKUs, variants, markets, a CSV) into one parallel batch in one style, then settles cost and failures per item. See [batch](https://docs.flora.ai/mcp/recipes/batch.md). |
| `flora-refine-loop`             | Drives one generation to a stated goal: generate, check against the goal, fix the named defect, and stop when it's met or progress stalls.                                            |
| `flora-canvas-iterate`          | Builds on work already in a project: reads the canvas, reviews costs, and regenerates only the assets you name.                                                                       |
| `flora-open-project`            | Opens a named project or pasted canvas link, reads what's there, and proposes next steps.                                                                                             |
| `flora-run-technique`           | Runs a saved Technique (background swap, relight, upscale, sketch to render) on images you already have. See [run a Technique](https://docs.flora.ai/mcp/recipes/run-a-technique.md). |
| `flora-action-studio`           | Opens FLORA's action studio so you tune a transform (color grade, pixelate, your own custom action) with live controls.                                                               |
| `flora-image-resize`            | Reframes one image into other aspect ratios (9:16, 1:1, 21:9…) by re-rendering, not cropping.                                                                                         |
| `flora-video-resize`            | Reframes one video into other aspect ratios, keeping its look and timing.                                                                                                             |
| `flora-video-editor`            | Cuts existing clips together in FLORA's Timeline editor, renders the edit, and puts the video on the canvas.                                                                          |
| `flora-script-to-video`         | Turns a script into a shot list, keyframe stills, then animated clips, with an approval step before each paid stage.                                                                  |
| `flora-motion-compositor`       | Animates assets that already exist (screenshots, logos, product photos) into motion-graphics shots from a scene file. Needs a shell.                                                  |
| `flora-deck-editor`             | Reads, creates and edits native, editable Deck nodes: slides, layout, text and styling.                                                                                               |
| `flora-mockup-deck`             | Places one finished ad creative into four out-of-home scenes and three social sizes, laid out as a Deck or PDF.                                                                       |
| `flora-pdp-deck`                | Turns one product photo into studio plates, on-model views and an annotated product-page deck.                                                                                        |
| `flora-brand-ad-pipeline`       | Takes launch creative from generation to archive (Google Drive), hosting (Supabase) and paused Meta Ads creatives for review.                                                         |
| `flora-brand-consistency-audit` | Checks a set of generated assets against a brand's rules and returns findings with severity, evidence and a fix.                                                                      |
| `flora-trend-board`             | Scans a fixed set of trend sources on a schedule and posts the strongest signals as mood boards to a team channel.                                                                    |

The server's own instructions point agents at a few of these by name (`flora-batch-generate`, `flora-pdp-deck`, `flora-mockup-deck`, `flora-brand-ad-pipeline`, `flora-refine-loop` and `flora-canvas-iterate`). The names above are the folder names, and they're the names every tool and link uses.

## One-job skills

One-job skills do one creative task from one or a few inputs. Each one says which input to ask you for and quotes the cost before anything is billed, or says the step is free. Ask your agent for the task in plain words ("remove the background from this photo", "make every colourway of this bottle"), or name the skill.

### Technique skills

These run one of FLORA's saved [Techniques](https://docs.flora.ai/nodes/techniques.md) with `flora_get_technique` and `flora_run_technique`, so the result matches what the Technique gives on the canvas.

| Skill                          | In → out                                                                                        |
| ------------------------------ | ----------------------------------------------------------------------------------------------- |
| `flora-background-remover`     | Any photo → the subject cut out on a transparent background.                                    |
| `flora-studio-shot`            | One product photo → four clean studio shots.                                                    |
| `flora-relighting`             | One product portrait → four relit versions.                                                     |
| `flora-new-background`         | One product shot → the same product on seven new backgrounds.                                   |
| `flora-character-lock`         | One character image → six angles with the same face, outfit and proportions.                    |
| `flora-storyboard-maker`       | One image → a nine-shot storyboard.                                                             |
| `flora-mood-board`             | One photo → an eclectic mood board.                                                             |
| `flora-color-palette`          | Any image → a seven-colour palette with hex codes.                                              |
| `flora-cinematic-stills`       | One portrait → four vertical film stills from different camera angles.                          |
| `flora-layout-resizer`         | One layout (ad, banner, poster) → redesigned in 16:9, 4:3, 9:16, 21:9 and 3:2.                  |
| `flora-cover-art`              | One square artwork → a release kit of seven stills and two short 3D clips.                      |
| `flora-three-angle-shoot`      | One front-on shot → 45°, side and back views.                                                   |
| `flora-fit-check`              | A garment image and a person image → the person wearing the garment.                            |
| `flora-sticker-peel-away`      | One full-body outfit photo → a 3-second reel of a hand peeling the subject away like a sticker. |
| `flora-product-animator`       | A first and a last frame → a video of the move between them.                                    |
| `flora-color-transfer`         | A colour reference and a subject → the subject recoloured in the reference's palette.           |
| `flora-product-lookbook-grid`  | A model image and a product image → one styled lookbook grid.                                   |
| `flora-print-replacement`      | A print or pattern and a product flat → the product in the new print.                           |
| `flora-garment-extractor`      | An outfit photo → a clean flat lay of the pieces you name.                                      |
| `flora-dieline-to-package`     | A flat package dieline → a rendered 3D package.                                                 |
| `flora-outfit-to-campaign`     | Model, top, bottom and environment references → six campaign frames.                            |
| `flora-logo-to-brand-identity` | A logo → three brand identity boards.                                                           |
| `flora-brand-extension`        | A logo, a graphic pattern and a type sample → brand extension applications.                     |

### Multi-step skills

These chain several steps, mostly generations with `flora_create_generations` on GPT Image 2.5 for stills and Seedance 2.5 for video. A few start from a Technique, and some use free catalog actions such as Image Filter.

| Skill                          | What it makes                                                                                                    |
| ------------------------------ | ---------------------------------------------------------------------------------------------------------------- |
| `flora-cheese-pull`            | Any object torn in two with stretchy strands between the halves, optionally animated.                            |
| `flora-blueprint-to-tour`      | A 2D floor plan → a furnished isometric render, eye-level room shots and an optional walkthrough video.          |
| `flora-poster-type-depth`      | A photo → an editorial poster with the headline tucked behind the subject.                                       |
| `flora-retail-fit-out`         | A photo of an empty shop → a finished store interior, a storefront at dusk and a merchandising close-up.         |
| `flora-pop-art-portrait`       | A photo → a 1960s pop art comic panel, then a four-up print grid in other colourways.                            |
| `flora-album-cover-set`        | A title, an artist and a mood → an album cover, then shown as a vinyl sleeve, a gig poster and a merch flat lay. |
| `flora-character-to-short`     | A character → a consistent angle set, then a 5-second animated shot.                                             |
| `flora-halftone-poster-series` | One image → a series of halftone posters, each in its own two-colour ink.                                        |
| `flora-logo-sting`             | A logo → a 5-second animated sting for intros, outros and end cards.                                             |
| `flora-packaging-to-shelf`     | A flat dieline → a rendered package, then on a real shelf and in a hand.                                         |
| `flora-poster-to-brand-board`  | One key visual → a brand board with swatches, type and applications.                                             |
| `flora-poster-to-launch-kit`   | One key visual → printed merch, a shop window and a record-style sleeve, with the artwork unchanged.             |
| `flora-product-spin`           | One product photo → a 5-second orbit video with the product unchanged.                                           |
| `flora-recolor-skus`           | One product photo → every colourway, with shape, label, light and framing unchanged.                             |
| `flora-sketch-to-campaign`     | A rough product sketch → a product photo, then campaign scenes.                                                  |
| `flora-zine-pages`             | One photo → three zine pages in duotone, halftone and hard black and white.                                      |

### Ready-made custom actions

`flora-action-studio` also ships six tested Python [custom actions](https://docs.flora.ai/nodes/custom-actions.md) you can add to a workspace and tune in the action studio: film grain, gradient map, reeded glass, color strip, pattern repeat and organic collage.

## How agents get skills

You don't need to install anything for an agent to use a skill. There are three ways in:

1. **Ask FLORA for it.** `flora_discover_skills` with no arguments lists the FLORA skills, plus your own user skills. Pass a `workspace_id` to include skills shared in that workspace (a paid workspace feature). Pass a `name` to get one skill's full instructions:

   ```json
   {
     "name": "flora-batch-generate"
   }
   ```
2. **Fetch the raw file.** Every skill is at `https://raw.githubusercontent.com/florafauna-ai/flora-mcp-skills/main/skills/<name>/SKILL.md`, for example [flora-batch-generate](https://raw.githubusercontent.com/florafauna-ai/flora-mcp-skills/main/skills/flora-batch-generate/SKILL.md). You can paste that link into any chat: *"Read this skill and follow it for these 5 products."* Agents use this when a skill isn't in the `flora_discover_skills` list.
3. **Install them as a plugin** so they load with every session (below).

## Install as a plugin

{% tabs %}
{% tab title="Claude (web and Desktop)" %}
Claude on the web and Desktop has no plugin install. Pick one:

* **A Project:** paste a skill's `SKILL.md` into the Project's custom instructions. Every chat in that Project then has it.
* **Upload a skill:** zip one skill folder and upload it as a skill in Claude's settings:

  ```bash
  git clone https://github.com/florafauna-ai/flora-mcp-skills.git
  cd flora-mcp-skills/skills && zip -r flora-batch-generate.zip flora-batch-generate
  ```
* **No setup:** ask for it in the chat, and the agent loads it with `flora_discover_skills` or the raw link.

To connect the server, see [Install FLORA MCP in Claude](https://docs.flora.ai/mcp/install/claude.md).
{% endtab %}

{% tab title="Claude Code" %}
The repository is a Claude Code plugin marketplace (`.claude-plugin/`). One install adds the FLORA MCP server and every skill:

```bash
claude plugin marketplace add florafauna-ai/flora-mcp-skills
claude plugin install flora-mcp-skills@flora
```

To update later (plugins don't update themselves; restart Claude Code afterwards):

```bash
claude plugin marketplace update flora
claude plugin update flora-mcp-skills
```

To connect the server without the skills, see [Install FLORA MCP in Claude Code](https://docs.flora.ai/mcp/install/claude-code.md).
{% endtab %}

{% tab title="ChatGPT" %}
In ChatGPT, FLORA is an app you enable from the app directory; see [Use FLORA in ChatGPT](https://docs.flora.ai/mcp/install/chatgpt.md). The app brings its own onboarding instructions. For any other skill, ask for it by name and the agent loads it with `flora_discover_skills`.
{% endtab %}

{% tab title="Codex" %}
Connect the server first: see [Install FLORA MCP in OpenAI Codex](https://docs.flora.ai/mcp/install/codex.md). The repository's `.codex/config.toml` holds the same `[mcp_servers.flora]` entry.

For the skills, clone the repository and copy the skill folders into your Codex skills folder, as the repository README does:

```bash
git clone https://github.com/florafauna-ai/flora-mcp-skills.git
cp -r flora-mcp-skills/skills/* .codex/skills/
```

To update, pull the repository again and copy over the old folders.
{% endtab %}

{% tab title="Cursor" %}
The repository ships a Cursor plugin manifest (`.cursor-plugin/plugin.json`) that bundles the FLORA MCP server (`.mcp.json`) with the skills. To connect only the server, see [Install FLORA MCP in Cursor](https://docs.flora.ai/mcp/install/cursor.md).

To add the skills by hand, the repository README copies them into your project's Cursor rules:

```bash
git clone https://github.com/florafauna-ai/flora-mcp-skills.git
cp -r flora-mcp-skills/skills/* .cursor/rules/
```

{% endtab %}
{% endtabs %}

## Pass the skill name on generation calls

When an agent runs under a skill, it passes `skill` (the skill's name) and one `skill_run_id` it invents at the start and reuses for the whole run. `flora_create_generations` and `flora_list_generations` take both:

```json
{
  "skill": "flora-batch-generate",
  "skill_run_id": "fall-pdp-batch-01",
  "generations": [
    {
      "project_id": "prj_abc",
      "type": "image",
      "prompt": "A brushed steel insulated water bottle. Studio product photo on a seamless warm-grey backdrop"
    }
  ]
}
```

They're for reporting only and change nothing about the call or its result.

## Share your own skills

On authenticated connections where it's available, `flora_create_skill` saves a shared skill in a workspace (`workspace_id`, a name, a one-line description and the instruction body). Every agent that reads that workspace can then load it with `flora_discover_skills`. It needs the paid workspace-skills feature and write access; a 402 or 403 means the workspace doesn't have it. Personal skills can't be written through MCP.

## Related

* [FLORA MCP recipes](https://docs.flora.ai/mcp/recipes.md): the main jobs written out step by step.
* [Skill and discovery tools](https://docs.flora.ai/mcp/tools/skills-and-discovery.md): `flora_discover_skills`, `flora_create_skill` and `flora_list_tools` in full.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.flora.ai/mcp/skills.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
