All plugins

berrys

Turn one product photo into on-brand listing images, campaign visuals and short product videos through Berrys AI's hosted MCP server (Streamable HTTP at https://app.berrys.io/api/mcp, OAuth sign-in, no keys in config). The agent plans the work on Berrys' visual canvas — hero, lifestyle and detail shots for a storefront, coordinated image sets and copy variants for a campaign, or a storyboarded multi-shot clip that keeps the product identical across cuts by seeding each shot from the previous frame — and the results stay on a shared canvas the user can edit, undo or restore. Uses Berrys' curated multi-model image and video stack, so there is no provider key to configure. Bundles the Berrys workflow skill. Agent Plugins v1 package (plugin.json + mcp.json + skills/), configuration only. Disclosure — every tool call goes to Berrys' hosted server (app.berrys.io) and acts on the user's own Berrys account and credits; the tool set and the playbooks it loads are served by that server at runtime and are not covered by the pinned sha. Generation spends Berrys credits: the server quotes a price first and every paid generation requires explicit approval in the conversation before it runs. Nothing is published to a marketplace or storefront on the user's behalf — assets are produced on the canvas for the user to download or use. The server signs in with OAuth (Berrys account). The package marks the server untrusted, so Hermes asks before every tool the server does not mark read-only and refuses them in unattended runs.

BerryAILABSToolsv0.3.5Updated 1 hour ago

About

Turn one product photo into on-brand listing images, campaign visuals and short product videos through Berrys AI's hosted MCP server (Streamable HTTP at https://app.berrys.io/api/mcp, OAuth sign-in, no keys in config). The agent plans the work on Berrys' visual canvas — hero, lifestyle and detail shots for a storefront, coordinated image sets and copy variants for a campaign, or a storyboarded multi-shot clip that keeps the product identical across cuts by seeding each shot from the previous frame — and the results stay on a shared canvas the user can edit, undo or restore. Uses Berrys' curated multi-model image and video stack, so there is no provider key to configure. Bundles the Berrys workflow skill. Agent Plugins v1 package (plugin.json + mcp.json + skills/), configuration only. Disclosure — every tool call goes to Berrys' hosted server (app.berrys.io) and acts on the user's own Berrys account and credits; the tool set and the playbooks it loads are served by that server at runtime and are not covered by the pinned sha. Generation spends Berrys credits: the server quotes a price first and every paid generation requires explicit approval in the conversation before it runs. Nothing is published to a marketplace or storefront on the user's behalf — assets are produced on the canvas for the user to download or use. The server signs in with OAuth (Berrys account). The package marks the server untrusted, so Hermes asks before every tool the server does not mark read-only and refuses them in unattended runs.

Berrys AI — MCP Server

MCP Badge

Official Website | App & Sign Up | Client Setup Guides

Your agent turns product photos into on-brand listing images, ads, and short videos on one canvas.

Berrys connects to the AI agent you already use: Claude, Claude Code, ChatGPT, Codex, Cursor, Gemini CLI, or VS Code. You plan the work with your agent. Berrys builds it on your visual canvas and runs the generation, with your credits and your approval.

What it does

  • Listing production — one product photo becomes hero, lifestyle, and detail shots for your storefront.
  • Campaign rollouts — hand over a brief; your agent plans the shot list, generates coordinated image sets and copy variants on a shared canvas.
  • Short product videos — storyboard a multi-shot clip and keep the product identical across cuts: the last frame of one shot seeds the next.
  • Live review — the agent shares a canvas link after each step so you can edit, undo, or restore.

Connect

The server is remote (Streamable HTTP + OAuth). There are no keys to copy — you sign in to Berrys in the browser on first connect.

Server URL: https://app.berrys.io/api/mcp

| Client | How to add |
|---|---|
| Claude (web/desktop) | Settings → Connectors → Add custom connector → paste the URL |
| Claude Code | claude mcp add --transport http berrys https://app.berrys.io/api/mcp |
| ChatGPT | Settings → Apps & Connectors → Advanced → Developer mode → Create connector with the URL |
| Codex | codex mcp add berrys --url https://app.berrys.io/api/mcp then codex mcp login berrys |
| Cursor | Add .mcp.json from this repo (or paste the URL in MCP settings) |
| VS Code | Add the mcpServers entry from .mcp.json |
| Gemini CLI | gemini mcp add --transport http berrys https://app.berrys.io/api/mcp |

First check (free): ask your agent "List my Berrys projects."

Repo layout

| Path | Purpose |
|---|---|
| .mcp.json | MCP server descriptor (auto-detected by Cursor Directory and Open Plugins clients) |
| skills/berrys/SKILL.md | Agent skill — how agents should drive Berrys (quotes, approvals, canvas workflows) |
| rules/berrys.mdc | Cursor rule for a consistent agent experience |

Spending credits

Building and browsing are free. Only generation spends Berrys credits. The agent shows an itemized quote first and waits for your explicit confirmation before submitting a charge.

Documentation

README from the reviewed commit 9ee3468; it updates when the author re-pins.

Developer Mode

Explore built-in Hermes toolsets and actions.

HermesAgent

Developer?

Create a Hermes extension for Desktop or agent workflows.

Developer Section