frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Show HN: Canary (YC) – Independent verification for AI code

https://www.runcanary.ai/
5•Visweshyc•44m ago
Hey HN, we are Aakash and Viswesh and we are building Canary (https://www.runcanary.ai/) - independent verification for AI code. Claude/Codex calls Canary with the changesets, intended behaviour and team knowledge. Canary then deploys agent swarms to investigate potential failures and test suspected runtime bugs in remote sandboxes.

To try it on your repository, paste this into your coding agent:

  Install the Canary CLI with npm i -g @runcanary/cli,
  then run canary skills and follow its instructions
  to onboard this repository.
Verification starts with what software is supposed to do and most importantly what it must never allow. This means investigating how inputs, permissions, state, timing, dependencies etc interact with each other. Intent is not always fully declared as well but many expectations are clear: private files should stay private, credentials should not leak, and retries should not create unintended duplicate effects.

We believe the future is a unified and independent verification system that starts with all those expectations and then chooses how to investigate each suspected failure. Source-only code reviews catches static issues in the implementation but even a clean review leaves a good chunk of behavioral only issues untested. Unit tests, integrations, E2E, static analysis, runtime experiments and formal verification are all means to establish that behavior thereby generating different kinds of evidence and guarantees.

This is why we believe a dedicated verification harness that can think and reason through all these modalities and invariants is necessary on top of general intelligence. The harness needs to start with the system’s intended behavior, develop a series of potential failure scenarios and choose how to investigate them. It’s sole functionality is to pressure test and challenge the assumptions behind a change, create the conditions needed to test suspected failures and assess what the resulting evidence establishes

How Canary works: it takes a cold snapshot of the codebase when called, combining the supplied intent and team knowledge with requirements, decisions, prior issues from tools like Notion, Linear. It can also route questions to you through the coding agents if anything is ambiguous.

Canary’s harness coordinates agent swarms by leveraging the different strengths across model families. It compares the code before and after, traces the effects through callers, dependencies, state transitions etc. and each suspected failure becomes a concrete scenario with an actor, state, trigger, outcomes and many more runtime states.,

For each suspected failure, Canary chooses the best way to provide evidence through methods like runtime verification, static analysis, unit, integration or sometimes even combination of these as necessary. The agent executes these checks in remote sandboxes by seeding data, configuring permissions, mocking dependencies and third party integrations and much more. Canary then returns these findings and supporting evidence back to the coding agents which then fixes these failures and requests reverifications against the failed scenarios.

To get started, give your coding agent this setup instruction and tell us what it caught and how we can do better.

  Install the Canary CLI with npm i -g @runcanary/cli,
  then run canary skills and follow its instructions
  to onboard this repository.
We are still pretty early in our journey and would love feedback on the product and how we can do better.

A perfect join algorithm? Answering queries in optimal time – Michael Arntzenius [video]

https://www.youtube.com/watch?v=ambisA2jegA
1•matt_d•1m ago•0 comments

Askholes and Low Effort AI Answers

https://dennisforbes.ca/blog/microblog/2026/09/askholes_and_low_effort_ai_answers/
1•llm_nerd•3m ago•0 comments

CrowdStrike: Frontier AI for Cybersecurity

https://strategyofsecurity.com/p/crowdstrike-frontier-ai-for-cybersecurity
1•mooreds•4m ago•0 comments

China reaches 5-minute EV charging

https://www.newsminimalist.com/articles/chinese-automakers-race-ahead-with-5-minute-ev-fast-charg...
1•russfink•4m ago•0 comments

Amazon Surprises by Actively Rehiring Former Employees

https://www.asiae.co.kr/en/article/2026092410112667195
1•mooreds•7m ago•0 comments

Show HN: Koi.rest – watch some fish and regain your balance

https://koi.rest
3•hxii•9m ago•0 comments

US Mobile CEO on Reddit: 1.1M customers, $1B+ on Warp, 5G SA, QCI 6, satellite

https://www.reddit.com/r/USMobile/comments/1w1kuvb/the_big_beautiful_update_11_million_customers_1b/
1•4d66ba06•10m ago•0 comments

Escaping Space: Part I

https://www.perplexity.ai/hub/blog/escaping-space-part-i
1•gmays•10m ago•0 comments

Vertical Farming: Reimagining Agriculture in an Urban Age

https://worldsensorium.com/vertical-farming-reimagining-agriculture-in-an-urban-age/
2•dnetesn•10m ago•0 comments

Remembering the Future: Blade Runner

https://blue-continuum.com/remembering-the-future-blade-runner
1•dnetesn•11m ago•0 comments

Dutch designer made DE9: Closer to the Edit into a playable web-based instrument

https://www.creativeboom.com/work/why-merijn-straathof-turned-a-landmark-techno-album-into-a-play...
1•ChrisArchitect•12m ago•1 comments

Becoming more than a meat proxy

https://codeplusconduct.substack.com/p/becoming-more-than-a-meat-proxy
2•mooreds•13m ago•0 comments

Meta's Muse page can't even pass basic privacy requirements

https://scanner.getprivisy.com/#/sales/96a75bdd-7033-4b66-9ad3-11041432abe2?token=eyJhbGciOiJIUzI...
2•patrickwdaly•15m ago•0 comments

New England Aquarium treating elderly penguin, 29, with acupuncture

https://www.neaq.org/about-us/press-room/press-releases/new-england-aquarium-treating-elderly-pen...
2•bookofjoe•15m ago•0 comments

Should we stop giving cyber threats fancy names?

https://www.techzine.eu/blogs/security/144469/should-we-stop-giving-cyber-threats-fancy-names/
1•VindtUMijTeLang•17m ago•0 comments

Why Buran Had Four Computers, Not Three

https://zatona.dev/blog/why-buran-had-four-computers
2•dzatona•20m ago•0 comments

Show HN: Public Browser MCP – +48% speed -33% token/session vs. AgentBrowser

https://github.com/Silbercue/public-browser/releases/tag/v3.0.0
1•silbercue•22m ago•0 comments

Education Consulting Industry Sucks

https://rivernova.vercel.app
2•roman9•23m ago•0 comments

Show HN: I rebuilt my company's old fire-station office as a Quake III map

https://sajarin.com/q3inc/
2•Sajarin•24m ago•0 comments

The New STOC Rules for the AI Era

https://blog.computationalcomplexity.org/2026/09/the-new-stoc-rules-for-ai-era.html
2•pcfwik•26m ago•0 comments

Show HN: Turn Your Voice into Art

https://reotoi.vercel.app
1•edimaudo•28m ago•0 comments

Empirical results fine-tuning π0.5 on a real manufacturing task

https://dream-machines.eu/blog/pi05-fine-tuning
1•mplappert•30m ago•0 comments

IR (2022)

https://github.com/dstogov/ir
1•lhmiles•33m ago•0 comments

Meta and the new VR glasses? Will Apple respond?

4•lathoa•34m ago•1 comments

China's new travel rules unsettle tech giants and talent

https://www.dw.com/en/china-technology-artificial-intelligence-expertise-capital-travel-rules/a-7...
10•billybuckwheat•35m ago•3 comments

Hail is a multibillion-dollar problem – and could get worse

https://www.hcn.org/issues/58-10/hail-is-a-multibillion-dollar-problem-and-could-get-worse/
4•littlexsparkee•35m ago•0 comments

Vibe Coding Production Kit – a production workflow for AI coding agents

https://github.com/Moeeryani/Vibe-Coding-Production-Kit
9•Moeeryani•39m ago•2 comments

Show HN: PipeFitter – Quick, Browser-Based, Parametric Pipe Connectors

https://mankyd.github.io/pipefitter/
2•mankyd•40m ago•0 comments

Show HN: Free WCAG accessibility checker – no signup, results in ~2 minutes

https://www.wcagrepair.com/tools/accessibility-checker
1•xLFCx•41m ago•0 comments

AI Help for Biological or Chemical Weapons

https://www.science.org/content/blog-post/ai-help-biological-or-chemical-weapons
2•EA-3167•43m ago•0 comments