Who it is forA user who is already working with an agent and needs to show it a visual problem, such as a layout, chart, or error screen.
The problemDescribing a visual issue in text takes effort and loses detail, and starting a new chat loses the earlier context.
The outcomeThe agent receives the image as native input inside the existing conversation, and the attachment stays with that conversation for later reference.
What you need first
- A Mac running macOS 13 or later on Apple Silicon
- Node.js 22 to 26 and npm 10 or later
- A Runta account with model access configured for your agents
- Git to clone the repository
Steps
- Open an existing agent conversation in Errand
- Attach an image to your message
- Send the follow-up message with a specific question about the image
- Check the agent's reply and the tool activity for how it used the image
- Reopen the conversation later to confirm the attachment is still there
Prompt for your agent
In the runta-dev/errand app, open the conversation for the agent <agent name> and attach the image <image path> to a follow-up asking <question>. Expected output: the agent's reply refers to content from the image. Verify that the attachment is still shown in the conversation after reopening Errand.
Caveat. This depends on an existing agent and a signed-in Runta account, and the README does not list which image formats are accepted.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for runta-dev/errand and "Send a screenshot to an agent as a follow-up"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer