The Agent Harness Hackathon: give AI models a license to act. August 24 to 30, 2026, with an NVIDIA DGX Spark and a Mac Mini among the prizes

The Agent Harness Hackathon

Give AI models a License to act

You get an agent working in an afternoon. Then you point it at something that matters and it can't reach your tools, can't run its own code safely, and can't be stopped before it does damage. Build one that can do all three, on TrueForge, TrueFoundry's open-source agent harness, and ship it the way you would at work: every change through a pull request, reviewed by Qodo before it merges.

Winners and projectsSee the prizes
Every submissionSubmit a project and you get a certificate of participation

Mission dossier

File TF-007

When
August 24–30, 2026. Monday 8 AM to Sunday 8 PM, London
Where
Take part ONLINE from anywhere, or join us in SF
Teams
Solo or up to 4 people
Prizes
$10,000 in prizes, including an NVIDIA DGX Spark and a Mac Mini, plus job interviews at TrueFoundry

Resources

Everything you need to build

The pages worth keeping open while you build, filed under the sponsor each of them belongs to. Stuck on something, or looking for a team? The WeMakeDevs Discord is the fastest way to an answer.

Read first

Getting Started Guide For The Agent Harness Hackathon

What a harness is and why an agent needs one, the seven steps to a first agent on TrueForge, what to build, and how Qodo fits into the week. Written by us, for this week.

Read the guide

Main sponsor · Agent harness

TrueFoundry

TrueForge is the harness every submission runs on. It is open source and it runs on your own machine, so everything here is something you can act on tonight.

Watch first

Main sponsor · Code review

Qodo

Qodo reviews the code you write around the harness, and a project has to show that it did to be considered for the Double-O and Q Branch tracks. Enough here to set it up, run a pull request through it, fix what it flags, and put the proof in your README.

If Qodo does not respond Confirm that the GitHub app can access the repository and that the repository is active in the Qodo portal, then comment /agentic_review on the pull request. If you do not have permission to install it, ask the repository owner or organisation admin to complete the one-time setup.

Fix the findings with a coding agent

Qodo's findings can be worked through by Codex or any other coding agent that supports skills. From the repository root, on the pull-request branch:

  1. Install the Qodo Agent Skills once. In Codex, run /skills to confirm qodo-pr-resolver is available.

    npx skills add qodo-ai/qodo-skills/skills

  2. Invoke it. In another compatible coding agent, select the same installed skill with that agent's skill command.

    $qodo-pr-resolver Resolve the Qodo findings for the PR on my current branch. Show the issues first, then let me review each one or auto-fix all.

  3. Review the proposed fixes, run the repository's tests, and approve or defer each finding with a reason. The skill can create fix commits, reply to Qodo's inline threads, and post a Qodo Fix Summary on the PR.

  4. Push the fixes. GitHub shows the fix commits, thread replies, and Fix Summary as visible remediation evidence. The push should trigger a follow-up review; if it does not, comment /agentic_review. Repeat until the review reflects the final code.

The loop: resolver, code change, tests, push, GitHub evidence, Qodo re-review.

Qodo Agent Skills

Tutorials