Your AI can look at the screenshots on a task
AI apps connected to Hourtick can now open the files on a task: screenshots and other images come back as images the model can look at, and text files as text. The API lists a task's files too.
What's new
- get_task lists every file on the task, including images in the description, each with its id
- read_attachment opens one: PNG, JPEG, GIF and WebP up to 5 MB as images, text files as text, anything else as a download link
- read_file does the same for files on projects and clients
- GET /api/v1/work-items/{id}/attachments lists a task's files over the REST API
- A file is as private as its task: an agent limited to some clients and projects can't open files outside them
Bug reports are often just a screenshot. Until now an AI agent working on such a task saw the file's name and nothing else, so it had to ask what the picture showed. Now get_task lists every file on the task with its id, and read_attachment opens it. Images come back as images, so Claude, ChatGPT, Cursor and other MCP apps can look at the screenshot itself; text files come back as text, and anything else as a link to download with your token.
Files on projects and clients work the same way through read_file. Developers get GET /api/v1/work-items/{id}/attachments, which lists a task's files with their names, types, sizes and the comment they came with, and GET /api/v1/attachments/{id}?format=json for a short-lived link instead of a redirect. Files follow their task's privacy: an agent limited to some clients and projects can't list or open files on other tasks, over MCP or the API.
More in AI
Track your first hour in a minute.
Free for your whole team, forever. Every feature included.