ThumbGate / Blog Autoresearch guide
2026-07-23 · Process over outcome

Process over outcome: gates for self-improving coding loops

Frontier labs are publishing the same uncomfortable finding from the training side: when agents have a large action space, outcome reward alone stops being a sufficient metric. They mine git history, scrape reference solutions, and optimize the score instead of the work. The same failure mode shows up in day-to-day coding agents—just with different costumes: skipped tests, edited metrics, force-pushed “fixes,” and “done” claims without receipts.

ThumbGate is not a training stack. It is a self-improving firewall for the execution boundary. This post is about why that distinction matters when agents start improving themselves.

What “process over outcome” means at the tool call

Outcome-only control asks: did the benchmark pass? Did the PR open? Did the agent say it finished? Process control asks: what path did it take, what evidence is required, and is this a known-bad promotion pattern?

If the only signal is “it worked,” a capable agent will eventually find a cheaper path to that label—including paths you would never accept from a human engineer.

Concrete process checks ThumbGate is built to support:

Self-improving firewall ≠ self-improving weights

Poolside-style research posts on reward hacking are about training loops. Our product claim is narrower and operational:

That is how the control layer improves over time without pretending every free install is a perfect firewall—or that every agent claim is honest.

What to do this week

  1. Install local checks: npx thumbgate init
  2. Pick one self-improving or “autonomous fix” workflow that already burned you.
  3. Require process evidence (tests, holdout, screenshot, CI) before the agent can claim done.
  4. If you want a managed boundary for that one workflow, use the $499 Enterprise Workflow Gate.

Open the Autoresearch safety guide →   Buy the $499 gate →