Skip to main content
Everything the agent reads or makes lands in a context library. A skill is a written workflow the agent follows. Together they are how one good session becomes repeatable work for the whole team.

​
The context library

The library has three levels. You see them in the Context panel of every session. The library is a real file system. The agent creates folders, moves files and reads what you drop in. Files have no size limit; large files upload straight to storage when they are saved.
Put brand guidelines, logos and product sheets in Workspace Context once. From then on, every session in the workspace starts with that context, and the agent does not ask for the files again.

​
Add files

  • Drag and drop files or folders into the chat composer or the Files panel.
  • Paste an image or a link. The agent opens links and reads the page as text.
  • Ask the agent to save something: “Save the approved stills in Workspace Context under Campaigns/Spring.”

​
Preview and download

The Files panel previews images, video, PDF, markdown, HTML and text. Download a single file, or ask the agent to zip a folder.

​
Memory

The agent remembers preferences you state in chat, such as “always export in 9:16 too” or “never use serif type for us”. Memory is stored in My Context and is personal. Ask the agent to forget an item to remove it.

​
Skills

A skill is a written workflow the agent follows step by step. Krea ships a set of built-in skills, and you can save your own.

​
Built-in skills

Built-in skills cover the common jobs, for example:
  • Cinematic Stills, the generalist default for photographic imagery.
  • Product Photography, Fashion Photography, Product Packaging Design, Poster Design, Static Ads, TV Show Thumbnails, Abstract Photography, Art Film Stills.
  • Krea Marketing, with workflows from a full ad campaign and UGC video ads to marketplace cards and Meta Ads performance creative.
  • Faceless Video for narrator-led explainers, and Krea Motion for Seedance 2.5 reveal films and product loops.
  • Frontend Design for standalone HTML pages the agent can open in the side panel.
Name a skill in your brief to use it. The agent also picks a fitting skill on its own when the brief matches one.

​
Save your own skill

1

Finish a job the way you like it

Run the workflow once. Steer the agent until the output matches what you want.
2

Ask the agent to save it

Write: “Save this process as a skill called Launch Teaser.” The agent writes the steps, the models it used and the checks it ran into a skill in your library.
3

Run it again

In a new session, name the skill: “Use the Launch Teaser skill on the attached product.” Edit the skill text at any time in the Skills panel.
You can also write a skill as a markdown file in another tool and import it.

​
Integrations

Under Integrations you connect external tools through OAuth: Slack, Notion, Linear, Airtable, Webflow, Pinterest, Meta Ads, Google Drive, Google Docs, Google Sheets, Gmail, Dropbox, OneDrive, GitHub, WhatsApp Business, YouTube, Supabase, TikTok Ads, ManyChat, Frame.io and Figma. Each tool has an Allow or Ask permission, so you decide which actions the agent may take without asking first.

​
Routines and custom agents

  • Routines run a brief on a schedule. Each run starts a fresh session with the same context library. Routines are available on Enterprise.
  • Custom agents are named agents with their own instructions, skills and integrations, for example a “Product Photo Agent” for one brand. Custom agents are available on Enterprise.