frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Ask HN: Is there an emerging pattern for NEGATIVESKILLS.md files?

3•mehmetoguzderin•9h ago
I am increasingly hitting cases where agents pick something from skills in part due to regex-based reading, since putting them all in context would be less efficient, leave out the negative, and do something exactly opposite of what is intended. Just like negative prompts for diffusion models, I feel like a negative-skills file listing explicit antipatterns and bad versions could enable harnesses with fewer graph nodes to perform better.

I might be missing some developments, but is there an accepted/shaping way to do this with recent models?

Comments

MrCoffee7•8h ago
This paper wasn't for SKILLS per se, but for LLMs, and suggests that LLMs can't follow too many constraints at one time very well https://arxiv.org/abs/2608.12426 . Other papers suggest that LLMs show more information sparsity in the results with negative constraints.
mehmetoguzderin•8h ago
Good one; reading this briefly took me back to the days of “Needle in a Haystack” being super challenging for LLMs. Maybe there needs to be a benchmark of “Rule in a Haystack” (similar to information one, testing not only independent rules also the ones that need hops) to clarify model performance regarding this. Thank you for the resource.

Ask HN: What are you working on? (September 2026)

131•david927•12h ago•365 comments

Ask HN: Would a startup for young creatives who reject AI be feasible?

9•mondobe•7h ago•13 comments

System 3 Thinking

2•rexossbear•5h ago•0 comments

Ask HN: What are some privacy/security tools you wish exist, but don't yet?

5•ToriTech•5h ago•6 comments

AI Boom / Bust – Boom Boom Pow

4•wwolfson97•6h ago•2 comments

Ask HN: How is AI actively a threat to humanity if it's only online?

5•CM30•6h ago•11 comments

Ask HN: Multi-agent workflows in production; Where people using 1000s of agents?

2•ramstar3000•7h ago•3 comments

Ask HN: Is there an emerging pattern for NEGATIVESKILLS.md files?

3•mehmetoguzderin•9h ago•2 comments

Ask HN: Should we start airgapping our personal devices?

7•emerongi•22h ago•6 comments

Ask HN: What default model do you use and why?

60•stikit•1d ago•101 comments

Ask HN: What would it look like if AI agents "took over the internet"?

4•scripper1•12h ago•11 comments

DietPi released a new version v10.7

3•stephanstsdietp•18h ago•0 comments

Ask HN: Career paths to consider if I am better at supporting than creating?

28•_override•18h ago•28 comments

Ask HN: Why haven't people vibe coded popular 80s word processors as TUI apps?

5•amichail•15h ago•3 comments

Ask HN: Did Google kill its enterprise workhorse model?

26•waldrews•2d ago•16 comments

Tell HN: OpenAI keeps re-enabling the 'allow training' setting

482•jacquesm•3d ago•188 comments

Ask HN: Can we please limit the AI news flood?

849•cromka•2d ago•392 comments

Reddit Now Blocking Firefox for Android with uBlock Origin?

25•chucksmash•1d ago•6 comments

Ask HN: What's a good "semi-dumb" phone/watch for kids?

8•funkaster•2d ago•12 comments

Ask HN: What is Y Combinator's position on Flock's standing today?

16•chirau•15h ago•21 comments

Tell HN: Claude Code appends new commit signature even if older settings is off

7•ctxc•1d ago•1 comments

OpenAI changed Navier-Stokes press release and Lean4 code on GH

9•rawland•2d ago•2 comments

Ask HN: For Enterprise coding agents, what's your company doing to control cost?

6•m0rde•1d ago•9 comments

Novgraph: Persistent Knowledge Graph for Codebases

3•Shourya_55•1d ago•2 comments

Ask HN: How are much smarter AI models made?

11•superasn•2d ago•5 comments

GoatCode – open-source terminal AI agent with provider failover

3•Arhan-w•1d ago•2 comments

You've reached the end!