Feed aggregator

Hi HN, I'm Ty. I built LUCID because I kept shipping bugs that my AI coding assistant hallucinated into existence.

Three independent papers have proven that LLM hallucination is mathematically inevitable (Xu et al. 2024, Banerjee et al. 2024, Karpowicz 2025). You can't train it away. You can't prompt it away. So I built a verification layer instead.

How it works: LUCID extracts implicit claims from AI-generated code (e.g., "this function handles null input," "this query is injection-safe," "this handles concurrent access"), then uses a second, adversarial AI pass to verify each claim against the actual implementation. You get a report showing exactly what would have shipped to production without verification.

"But can't the verifier hallucinate too?" Yes -- and that's the right question. The benchmarks below were validated by running real test suites, not by trusting LUCID's judgment. The value is that structured claim extraction + adversarial verification catches bugs that a single generation pass misses. The architecture also supports swapping LLM verification for formal methods (SMT solvers, property-based testing) per claim type as those integrations mature.

Benchmarks:

- HumanEval: 86.6% baseline -> 100% pass@5 with LUCID (164/164 problems) - SWE-bench: 18.3% baseline -> 30.3% with LUCID (+65.5%) - Both benchmarks were validated by running actual test suites, not by LLM judgment - LLM-as-judge actually performs worse at higher k values -- it hallucinates false positives

Three ways to use it:

1. MCP Server (Claude Code, Cursor, Windsurf) -- one config line, verification as a native tool 2. GitHub Action -- automated verification on every PR with inline comments 3. CLI -- npx lucid verify --repo /path/to/code

Free tier: 100 verifications/month. Get a key at https://trylucid.dev

Code: https://github.com/gtsbahamas/hallucination-reversing-system Paper: https://doi.org/10.5281/zenodo.18522644 Dashboard: https://trylucid.dev

Comments URL: https://news.ycombinator.com/item?id=47011695

Points: 2

# Comments: 0

Article URL: https://www.heise.de/en/news/German-language-Wikipedia-considers-comprehensive-AI-ban-11175670.html

Comments URL: https://news.ycombinator.com/item?id=47011683

Points: 1

# Comments: 0

Article URL: https://lwn.net/SubscriberLink/1057561/bddc1e61152fadf6/

Comments URL: https://news.ycombinator.com/item?id=47011676

Points: 1

# Comments: 0

Article URL: https://challengermap.ca/

Comments URL: https://news.ycombinator.com/item?id=47011651

Points: 1

# Comments: 0

Most “AI + browser” setups still bolt MCP tools onto Playwright and hope for the best, so every click dumps full DOMs, accessibility trees, and logs into the model.

That burns tokens, collapses context, and makes long sessions unreliable.

Meanwhile, default Playwright reports start to struggle once you have more than a few dozen e2e tests, so teams drown in HTML reports and flaky failures instead of clear patterns.

The insights at https://testdino.com/blog/playwright-cli/ explores how Microsoft’s playwright-cli keeps browser state external, returns only compact element references and YAML flows, and works with normal npx playwright test plus smarter reporting, so both agents and humans stay fast, cost aware, and predictable.

Comments URL: https://news.ycombinator.com/item?id=47011649

Points: 1

# Comments: 0

I built an API that takes a product name, scrapes reviews from YouTube and Reddit, and returns structured sentiment analysis in a single JSON response. Live demo (no signup): https://reviewstack.vercel.app/demo

The response includes a normalized score (1-10), a plain-text summary, pros/cons lists, recurring themes with sentiment, and source attribution linking back to the original content.

The AI layer uses Claude by Anthropic. It reads the collected reviews and extracts structured data. The value is in not having to maintain scraping infrastructure, handle rate limits across platforms, or write your own extraction prompts.

Stack: Next.js API routes, Vercel for hosting, Stripe for billing, YouTube Data API + Reddit JSON endpoints for sourcing, Claude for analysis.

Pricing: free tier at 50 lookups/month, paid plans at $29/mo (500 lookups) and $79/mo (2,000 lookups). Solo/bootstrapped project.

Happy to answer questions about the scraping approach, accuracy, or anything else.

Comments URL: https://news.ycombinator.com/item?id=47011640

Points: 1

# Comments: 0

Listen for it for 7 days and you will be completely FREE

Comments URL: https://news.ycombinator.com/item?id=47011634

Points: 2

# Comments: 0

Article URL: https://chess-pulse-neon.vercel.app/

Comments URL: https://news.ycombinator.com/item?id=47011630

Points: 1

# Comments: 1

Article URL: https://ourworldindata.org/data-insights/chinas-adoption-of-industrial-robots-has-surged-over-the-past-decade

Comments URL: https://news.ycombinator.com/item?id=47011618

Points: 1

# Comments: 0

Article URL: https://www.backblaze.com/blog/backblaze-drive-stats-for-2025/

Comments URL: https://news.ycombinator.com/item?id=47011602

Points: 2

# Comments: 0

Article URL: https://abio.substack.com/p/dc-required-daycare-workers-to-get

Comments URL: https://news.ycombinator.com/item?id=47011592

Points: 1

# Comments: 0

Article URL: https://webkit.org/blog/17818/announcing-interop-2026/

Comments URL: https://news.ycombinator.com/item?id=47011581

Points: 1

# Comments: 0

Article URL: https://www.newscientist.com/article/2514600-why-exercise-isnt-much-help-if-you-are-trying-to-lose-weight/

Comments URL: https://news.ycombinator.com/item?id=47011568

Points: 1

# Comments: 1

sql-tap is a transparent proxy that captures SQL queries by parsing the PostgreSQL/MySQL wire protocol and displays them in a terminal UI. You can run EXPLAIN on any captured query. No application code changes needed — just change the port.

Comments URL: https://news.ycombinator.com/item?id=47011567

Points: 2

# Comments: 0

Article URL: https://www.youtube.com/watch?v=Z5GKnb4H_bM

Comments URL: https://news.ycombinator.com/item?id=47011555

Points: 1

# Comments: 0

Article URL: https://www.youtube.com/watch?v=SuTxpKggY30

Comments URL: https://news.ycombinator.com/item?id=47011549

Points: 1

# Comments: 0

Article URL: https://github.com/NationalSecurityAgency/ghidra

Comments URL: https://news.ycombinator.com/item?id=47011548

Points: 2

# Comments: 0

Hi HN, I built MicroVibe, a small web starter for people who want JSX + file-based routing + API routes, without pulling in a heavy framework.

What it does today:

- File-based routing (including dynamic and catch-all segments)

- API routes with consistent JSON error shape

- SSR by default, and `mode = "client"` per route when interactivity is needed

- Runtime module cache with file-change invalidation in local dev

Project goal: keep the runtime small and understandable so teams can iterate quickly and still reason about behavior.

I would really value feedback on:

1. Routing/API ergonomics

2. What was confusing or slow in your first 30 minutes using MicroVibe?

3. Where this should clearly differ from Next/Astro/Vite workflows

Comments URL: https://news.ycombinator.com/item?id=47011545

Points: 3

# Comments: 1

Article URL: https://twitter.com/merettm/status/2022517085193277874

Comments URL: https://news.ycombinator.com/item?id=47011542

Points: 1

# Comments: 0

Article URL: https://qualify.gauntletai.com/

Comments URL: https://news.ycombinator.com/item?id=47011541

Points: 1

# Comments: 0