The Agent Harness Hackathon: give AI models a license to act. August 22 to 30, 2026, with an NVIDIA DGX Spark and a Mac Mini among the prizes

Sponsored by TrueFoundry · Hosted by WeMakeDevs

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.

Register nowSee the prizes

Mission dossier

File TF-007

When
August 22–30, 2026
Where
Take part ONLINE from anywhere, or join us in SF
Teams
Solo or up to 4 people
Prizes
$10,000 in prizes, include an NVIDIA DGX Spark, a Mac Mini, and more

Rules

Everything you're held to

Read these before you start building. Registering for The Agent Harness Hackathon means agreeing to all of them, along with the WeMakeDevs Code of Conduct.

  1. You can take part online from anywhere in the world, and it costs nothing to enter. If you're in San Francisco on August 29, you can also spend the day building in the room.*

  2. You can participate solo or in a team of up to 4 members. Each participant may only be part of one team.

  3. Required technology: your agent must run on TrueForge, the open-source agent harness. A judge has to be able to see the harness doing real work rather than sitting under a thin wrapper around a model call.

  4. Beyond that the challenge is open-ended. Build a developer tool, an internal assistant, a research desk, an incident responder, a data pipeline, or anything else worth handing to an agent, in any domain you like.

  5. Your submission must be open source. Judges have to be able to read the code and run it.

  6. Anything your agent touches has to be yours to touch. Connect tools, data, and accounts you own or have permission to use, and keep private, personal, and login-protected information out of your repo and your demo.

  7. The project has to be built during the hackathon. You may discuss ideas, take notes, plan the architecture, or prepare diagrams beforehand, but the coding and design work itself has to happen between August 22 and the deadline.

  8. You may use frameworks, open-source libraries, public APIs, templates, third-party tools, and publicly available assets. The original work completed during the hackathon is what gets judged.

  9. Every submission must include:

    • A public source-code repository
    • A clear README with setup steps
    • A demo video of about three minutes showing the agent working
    • A short write-up of what the agent does and how it uses TrueForge
    • A link to your blog post, if you're entering that prize

  10. Submissions close on August 30 at 8:00 PM San Francisco time, and the schedule page shows that deadline in your own timezone.

  11. AI coding assistants are allowed, but their use must be disclosed.

  12. Participants must understand the submitted code and be able to explain the agent, the project architecture, and the technical decisions behind it.

  13. Projects that are entirely generated using AI without meaningful participant contribution, verification, or technical understanding may be rejected.

  14. There are two judged tracks, and each awards its prize to the winning team. Every submission is considered for both, but one team can only take one of them.**

  15. The blog post prize goes to one writer: publish your write-up anywhere you like and add the link to your submission. Swag goes to ten participants who share their build publicly and tag WeMakeDevs and TrueFoundry.

  16. The repo draw is made at random after submissions close from everyone who has starred TrueForge on GitHub, one entry each, and the draw is final.

  17. Any intellectual property developed during the hackathon belongs to the participant or team that created it. Teams are encouraged to agree internally on ownership before submitting.

  18. Treat participants, organisers, sponsors, speakers, judges, and community members with respect.

  19. Harassment, discrimination, plagiarism, or attempts to manipulate the judging process will result in disqualification.

  20. Failure to follow these rules or the WeMakeDevs Code of Conduct may result in disqualification.

  • *The live day has limited space and takes a registration of its own on Luma. Everyone who turns up gets $50 in OpenAI credits; online participants bring their own model API key.
  • **Running your pull requests through Qodo is required to win the Best Code Quality track. Nothing else in the hackathon depends on it.

Disclaimer: The Agent Harness Hackathon is an independent developer hackathon organised by WeMakeDevs in collaboration with TrueFoundry. It is not affiliated with, endorsed by, or associated with the James Bond films or novels, Eon Productions, Danjaq, Metro-Goldwyn-Mayer, or any of their rights holders. The theme is used purely for creative purposes.