Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. ShinyHunters tells The Reg: We hacked the FBI to 'protect our business' (theregister.com)
    1comments
  2. Canadian language school files for bankruptcy, leaving students stranded (cbc.ca)
    —discuss
  3. Created an online job aggregator platform and I need ideas (labelingjobs.net)
    —discuss
  4. Show HN: Paper-docx – agent-native Python-docx fork with 78% fewer DOCX failures (github.com/paper-instruments)
    —discuss
  5. Beef Grading Shields (usda.gov)
    —discuss
  6. Show HN: CK3DNA – searchable CK3 character DNA and coat of arms codes (ck3dna.com)
    1comments
  7. Approaching a 10 Second Linux Kernel Build (phoronix.com)
    —discuss
  8. Show HN: Digitron – a virtual analog synth and sequencer (apps.apple.com)
    —discuss
  9. AI Infra Is Nothing Like the Classic Cloud Infra (ramansharma.substack.com)
    —discuss
  10. Lobsters: Rename vibecoding to LLMs (Greasemonkey script) (greasyfork.org)
    —discuss
  11. Bringing HTTP's caching rules to DuckDB functions with VGI (query.farm)
    —discuss
  12. No country controls every layer: The hidden AI systems reshaping global power (abc.net.au)
    —discuss
  13. Fedora considers replacing LibreOffice with Collabora Office (digitalescapetools.com)
    —discuss
  14. Trump administration gutted data collection on US education (theguardian.com)
    —discuss
  15. Code is the future and present of hardware engineering (charlesazam.com)
    —discuss
  16. Zelensky says Russia has widened attacks to hit Ukraine's data centres (bbc.com)
    —discuss
  17. Show HN: Get a .has-a.computer sub-domain for your website (has-a.computer)
    —discuss
  18. Poor Privacy Practices Leaves Revolut Customers Vulnerable (2020) (safereddit.com)
    1comments
  19. Drex: Open jev-like claims win on decision index (nace.ai)
    —discuss
  20. Gaming is expensive, ya, I'm Good (yaimgood.bearblog.dev)
    —discuss
  21. Show HN: Obsidian Plugin for OpenModelica (community.obsidian.md)
    —discuss
  22. New photos show damage at U.S. positions caused by Iranian attacks (cbsnews.com)
    —discuss
  23. Ask HN: Which model do you use for work?
    2comments
  24. Microsoft disrupts AI-assisted platform that compromised 12,000 accounts (arstechnica.com)
    —discuss
  25. Resell sneakers until you're a billionaire on OBLOCK (mattbusel.itch.io)
    1comments
  26. Cambridge Analytica 2.0 – When the AI Assistant Becomes a Intelligence Graph (europa.eu)
    —discuss
  27. Levyra: Open-source music player for Android and Windows (github.com/luc4n3x)
    —discuss
  28. Best TV shows of the 21st century (nytimes.com)
    —discuss
  29. Treat Agents Like Direct Reports (toppingdesign.com)
    —discuss
  30. Sherline Auction of Its Manufacturing Machines (bidspotter.com)
    —discuss

Show HN: Stop shipping AI slop, Executor makes agents prove it works

1 pointsby 1h agogithub.com
1 comments
1h agoHN ↗

AI slop in code has a pretty recognizable shape: code that was never actually verified, merged because the agent sounded confident.

“Done, all tests pass.”

Except they don’t.

It’s usually not malicious. The agent just can’t reliably tell the difference between what it intended to do and what it actually did. So the fix isn’t necessarily a better prompt. It’s taking verification authority away from the agent itself.

That’s what The Executor does.

It’s a set of Markdown skills + POSIX bash scripts that takes a feature through:

intake → spec → plan → execution → review → verification → handoff

The important part is that state transitions are enforced by scripts that fail with a non-zero exit code, rather than by instructions the agent is expected to follow.

It works with Claude Code, Codex, or pretty much any harness that can read skill files. No daemon. No SaaS.

How it makes agents prove their work:

Reviewers see the diff, not the implementer’s claims.

The implementation agent’s “looks good” report is not evidence. Review verdicts are stored as files, and a FAIL verdict physically blocks exec-branch merge. The merge command simply refuses to run.

Every finding has to point somewhere specific.

There’s one ID namespace per initiative, and every finding references the exact spec requirement it violates.

For example:

INIT-0004-P01-T03-R02 fails INIT-0004-SPEC-01-R07

No vague “this doesn’t look right” findings with nowhere to attach them.

Verification happens from scratch.

Each spec criterion gets its verification command run against the current commit, producing one of:

PROVEN / FAILED / NOT-RUN / UNAVAILABLE

And this matters: a single NOT-RUN means you cannot call the feature “complete.” Nothing gets upgraded by inference.

Fix loops don’t continue forever.

At round 4, the process escalates to a different model. Hitting the cap requires an explicit recorded ruling, so findings can’t just disappear because the agent got tired of fixing them.

Context resets don’t reset reality.

After a context reset or model switch, the controller re-reads the ledger instead of trusting the model to remember what happened.

exec-run check audits the registry, ledger, and verdict files against each other and exits with 1 while naming the inconsistency.

There is a cost, though: ceremony and tokens.

This isn’t really for “change one line and ship.” It’s aimed at features that live for days and pass through multiple rounds of implementation, review, and verification.

MIT licensed.

Roast it. Where does this break down at your scale?