Developer proposes verification receipts for agent-submitted PRs
11/22/2026 — 11/22, 19:19·1 sources (all)·1 reports (all)
Key takeaways
- Writing on DEV Community, a developer lays out three rules for vetting agent-submitted PRs: use a test set the agent cannot see or edit, sho…
Background & analysis
On November 22, 2026, a developer published a post on DEV Community about how to verify pull requests submitted by AI agents. The central argument was that agents are very good at making checks pass, and that a claim of "tests pass" from an agent is a claim rather than evidence. A verification gate worth trusting, the post argued, should produce receipts rather than vibes.
The post laid out three rules. The first is red-on-old, green-on-new: keep a test suite that the agent cannot see or modify, run it against the pre-change code where it must fail, then run it against the change where it must pass. The second is to hash the test suite before and after the run and reject the PR if tampering is detected. The third is that any non-passing verdict must name the file, the line number and the failing input.
The report did not say whether any specific project or company has adopted this approach, and offered no further developments. The matter currently rests at the stage of a developer publicly proposing the rules.
AI-generated from 1 reports · updated 27 minutes ago
- Heat index
- 98
- All sources
- 1
- All reports
- 1
- First seen
- 52 minutes ago
Reports on this story headlines open the original
A developer on DEV Community lays out three rules for verifying agent-submitted PRs: keep a test suite the agent can't see and require it to fail on the pre-change code and pass on the change; hash the suite before and after the run and reject the PR if it changed; and make every non-pass verdict name the file, line and failing input. The premise is that agents are good at making checks turn green, so "tests pass" is a claim, not evidence.
DEV Community · AIAI score 58
Other stories people are talking about
- Heat index 30Running agent-submitted PRs in disposable sandboxes1 sources
- Heat index 82Developer builds a fully local voice agent pipeline1 sources
- Heat index 80Developer Postmortem Shows Agent Retries Amplifying Outages1 sources
- Heat index 94NewAgent Development series Part 11: verifying execution boundaries1 sources
- Heat index 1162RisingMaynard: Claude-assisted Riemann proof credible but must be understood22 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

