Grade how candidates work with AI, not just the code they ship.

Vibe coders produce passing tests. Good engineers guide the AI, understand what it built, and catch what it got wrong. Heedful records the whole session so you can see which one you're hiring.

Get early access

Step 1

Turn real work into a take-home exercise

Use our agent skill to turn any PR into a realistic take-home exercise for your next hire. Provide context in an interactive agent session and get a minimal reproduction of the problem alongside a grading guide that compares their solution to your team's.

→ taskforge on GitHub

Step 2

Share a link with the candidate

Upload the exercise into Heedful, set a time limit and AI budget, then send a link. The candidate gets a sandboxed VS Code environment they're already familiar with, Claude already set up. Heedful records every step and notifies you when they're done.

Candidate environment: VS Code in the browser with Claude Code running in the terminal

Step 3

Learn how they think

Heedful evaluates not just the final solution but every prompt, tool use, and line of code written during the session. You get an overview of how the candidate got there, how their solution compares to your team's, and a set of follow-up questions for the next interview step.

Heedful console: timeline view showing candidate's step-by-step AI collaboration

Get early access

Heedful is invitation-only right now. Leave your email and we'll reach out.

Request access →