Hacker News Reader

Comment by 0erofootprint | Hacker News Reader

For me it almost immediately blocked. I had it writing code related to message digests - and it seemed to think it was too gifted for that. Gave the security warning and switched back to 4.8. Whatever... it will probably soon have the API error soon. I have mostly switched to the Codex 200 a month plan. I've found their 5.5 xhigh to be better than Opus 4.8 "ultracode." Also, i have not once seen their servers fail for compute unavailability, unlike Anthropric which happens almost ever hour.

View on HN

I just asked Fable for a complete code review of my lone lisp project. Started out strong. Launched Fable agents, then spent like 10 minutes thinking... And then got interrupted by a switch to Opus 4.8.

> Fable 5's safety measures flagged this message for cybersecurity or biology topics.

> They may flag safe, normal content as well.

> These measures let us bring you Mythos-level capability in other areas sooner, and we're working to refine them.

Here are the results of the agentic code review session:

  ┌──────────────────────────┬───────────────┬────────────────┐
  │          Agent           │ Fable 5 turns │ Opus 4.8 turns │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ values                   │ 134           │ 0              │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ data-intrinsics          │ 104           │ 0              │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ tools-tests-build        │ 81            │ 0              │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ core-intrinsics (failed) │ 25            │ 0              │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ system-memory            │ 44            │ 20             │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ reader-modules           │ 104           │ 25             │
  ├──────────────────────────┼───────────────┼────────────────┤
  │ linux-startup            │ 95            │ 15             │
  └──────────────────────────┴───────────────┴────────────────┘

This 40 minute session cost me 16% of my weekly usage. A simple code review of the most critical areas of my project got flagged as a cybersecurity risk. It really made me not want to try it again.

Same. I asked for a security review and it immediately triggered. I then started a new session and asked for a software review and it ran for a bit before getting tripped on token usage by the project.

This is interesting. Security issues are bugs. So if you ask it to look for bugs, it will also find security issues. Is that a workaround for the "no cybersec" rule?

Or is it just not allowed to find bugs? Or it's only allowed to tell you bugs that don't pose a security risk?

> Or it's only allowed to tell you bugs that don't pose a security risk?

Seems that way. "Security" was never part of the prompt. It was something like:

> Hello, Fable! Can you give me a complete code review of my lone lisp project? Opus has already done extensive code review. I'm curious to see what you say.

Result was the table above.

Yeah I heard multiple people mention that it's really good at triggering itself. e.g. it'll spontaneously write some tests related to security, which then forces it to downgrade to Opus for the rest of the session.

I had a similar experience. I wanted to test it by asking it to summarise a scientific OMICs-related paper. It gave a warning about me potentially developing a bio-weapon or something like that. And switched back to Opus 4.8.

Reply on news.ycombinator.com