Scenario open-sources a set of Agent Skills for generative content in AI agents
XORit Editorial · Sources, AI assistance and corrections
Scenario has published a set of Agent Skills in an open repository. These are instructions an AI agent follows to create content through the service's MCP server: images, video, audio, textures, skyboxes, 3D assets and its own trained models. It is aimed at games, entertainment and other creative tasks.
The skills are built to the Agent Skills format, so they work in Claude Code, Cursor, Codex, Copilot and roughly seventy other agents that support that format. The full set installs with the command npx skills add scenario-labs/skills –skill "*"; the same command without the flag opens the installer menu.
Some skills reference neighbouring ones, so related pipelines, for example for video ads or a music video, are installed together: otherwise those references stay unfilled.
Using them requires a connected Scenario MCP server at https://mcp.scenario.com/mcp, with sign-in through an OAuth account or an API key. Setup details are gathered in the base scenario skill.
The README lists example tasks: making four potion icons in a consistent style with a transparent background, building a seamless brick texture and upscaling it to 2048 without breaking the tiling, and turning a sketch into a 3D prop with a preview before exporting to GLB.
Installation always pulls a fresh copy of the skills from the main branch; there is no version pinning. Changes become visible with releases, which also land in the CHANGELOG and in the service's changelog. Updates can be pulled with the command npx skills update or by reinstalling.
The repository is public, under the MIT licence, the scripts in the skills have tests, and pull requests are welcome. Formally, a skill is a SKILL.md file with procedural instructions that the agent loads on demand.
Translation editor: Салтанов А.
Source: github.com ↗