Hacker News: GustavHartz

Tell HN: OpenAI $500 ProMax plan listed in API

GustavHartz — Fri, 25 Sep 2026 08:12:24 +0000

OpenAI's API is showing the first signs in the API of a new 500$ ProMax plan

https://chatgpt.com/backend-anon/checkout_pricing_config/con...

Comments URL: https://news.ycombinator.com/item?id=49841605

Points: 29

# Comments: 51

New comment by GustavHartz in "Pi Security – Codex Security without all the bloat"

GustavHartz — Fri, 14 Aug 2026 13:52:47 +0000

OpenAI recently released Codex-Security. We simplified it based on what we use in our own products and put it in the PI harness. We will release data on DeepSeek v4 and GLM-5.3 performance soon, but so far it's probably the best open-source vulnerability detection tool out there right now.

The idea is the same as Codex Security.

1. Build a threat model 2. Launch a lot of probing agents looking into the security based on the threat model 3. Deduplication 4. Validation 5. Severity and likelihood calibration

We did a deep dive article here on it as well https://x.com/GustavHartz/status/2084926544800035266

Pi Security – Codex Security without all the bloat

GustavHartz — Fri, 14 Aug 2026 13:52:47 +0000

Article URL: https://github.com/Cecuro/open-security

Comments URL: https://news.ycombinator.com/item?id=49298705

Points: 4

# Comments: 1

Under the Hood of Codex Security

GustavHartz — Wed, 05 Aug 2026 09:04:12 +0000

Article URL: https://twitter.com/GustavHartz/status/2084926544800035266

Comments URL: https://news.ycombinator.com/item?id=49180335

Points: 5

# Comments: 2

New comment by GustavHartz in "Codex Security"

GustavHartz — Wed, 05 Aug 2026 09:02:38 +0000

TLDR on how it works: It's a small stack of skill files and some JS code that starts a large number of Codex sessions. They all get the prompt and the same scope with a limited set of tools. Not one prompt in the repo contains security advice, performance is obtained only through scaling the number of agents looking at the code

What 650k commits say about how crypto bugs change

GustavHartz — Mon, 06 Jul 2026 10:30:58 +0000

Article URL: https://twitter.com/GustavHartz/status/2073819470796128402

Comments URL: https://news.ycombinator.com/item?id=48802836

Points: 2

# Comments: 0

More agents are better than fine-tuned model for pen-testing

GustavHartz — Wed, 01 Jul 2026 12:25:59 +0000

Article URL: https://twitter.com/GustavHartz/status/2072294954404135275

Comments URL: https://news.ycombinator.com/item?id=48745599

Points: 1

# Comments: 0

New comment by GustavHartz in "Building effective pen-testing agents"

GustavHartz — Fri, 26 Jun 2026 08:16:30 +0000

This started as a response to the recent "you have to post-train a model to pen-test" Show HN — we don't think you need to, just makes life a bit easier.

Across 10K+ of our agent transcripts from benchmarking against OpenAI's EVMBench, we saw zero refusals. In the closed-frontier models, the refusal you hit is mostly a separate content classifier, or a system prompt, not so much the model itself. Breadth (more cheap agents) beats a bigger model, but it puts more requirements on context engineering

https://news.ycombinator.com/item?id=48609231

Building effective pen-testing agents

GustavHartz — Fri, 26 Jun 2026 08:16:30 +0000

Article URL: https://cecuro.ai/blog/building-effective-pen-testing-agents

Comments URL: https://news.ycombinator.com/item?id=48683859

Points: 5

# Comments: 1

New comment by GustavHartz in "In 92% of DeFi exploits AI security review flags underlying problem"

GustavHartz — Mon, 23 Feb 2026 12:17:31 +0000

Performance has gotten a lot better the last 6 months, at a level where we almost don't see it anymore at Cecuro.ai. PoC generation and multiple validation agents debating validity is the key differentiator. This is an ok paper on the topic https://arxiv.org/abs/2511.02780

In 92% of DeFi exploits AI security review flags underlying problem

GustavHartz — Sun, 22 Feb 2026 13:43:37 +0000

Article URL: https://www.coindesk.com/business/2026/02/20/specialized-ai-detects-92-of-real-world-defi-exploits

Comments URL: https://news.ycombinator.com/item?id=47110937

Points: 3

# Comments: 2

OAI: EVM Bench LLM Accuracy on Smart Contract Review and Pentesting

GustavHartz — Thu, 19 Feb 2026 08:47:51 +0000

Article URL: https://openai.com/index/introducing-evmbench/

Comments URL: https://news.ycombinator.com/item?id=47071489

Points: 2

# Comments: 0

New comment by GustavHartz in "AI agents find $4.6M in blockchain smart contract exploits"

GustavHartz — Sun, 07 Dec 2025 16:42:22 +0000

We've been working on this at cecuro.ai. When we test Sonnet 4.5 against real cyber security audit reports from the major firms on code that came out after the model was trained, it finds around 95% of the same bugs the auditors found. Also catches some medium severity stuff they missed. We find that you can't just point one model at a contract and expect good results though. Need to run multiple models with different prompts because they each have different blind spots. Still tricky to get working well and not cheap. Happy to share more if anyone's curious

Ask HN: Do you think MS Copilot will replace junior consultants?

GustavHartz — Sun, 13 Apr 2025 14:37:09 +0000

As a consultant, I spend way too much time buried in emails, teams, Slack threads, and random SharePoint folders trying to come up with answers to questions like: • What’s the current project status? • What is that XXX deck? • What is the revised timeline?

Junior consultants at my company usually get the job of building timelines, trackers, and endless update slides. I’m curious whether MS Copilot, due to the data access, could fill this role—parsing all comms, tracking deliverables, and giving you clean overviews, action items, and even draft decks.

Do you think Copilot will evolve into this kind of tool? Or is there still a gap here that a startup could fill? Curious if others are seeing the same pain, or already building around it.

Idea visualized: https://theo-consult-canvas.lovable.app/dashboard

Comments URL: https://news.ycombinator.com/item?id=43673131

Points: 1

# Comments: 0