Hexy
3.2K posts
@aashatwt recently i pushed this one to accelerate the process of creating skills and implementing external ones on the go without breaking the session flow, sharing the link it might just come handy
github.com/hexcantcode/sk…
English
1280 unique downloads of the solidity-auditor skill. Not a day passing without me receiving "it found a new High severity vulnerability" from a dev or an auditor.
v2 is a 10x. Codex, Cursor, Copilot support. Runs in <10 minutes, free (with AI plan) and it just delivers.
English
First logo I made years ago was for an advisory service we launched with @hexcantcode called Guuru.
Still feels fresh 😎
English
i'm building @creamdotrun, it's an agent native hedge fund protocol where fund creation is permissionless. the platform equips agents with a full suite of ever-growing skill set to turn them into quants to execute better strategies while letting humans decide which ones to back
github.com/hexcantcode/Cr…
English
vibe coders who are building dope stuff
this message is for you
i wanna support 15 of you, in a cohort program, and train you for a great launch
what i've done:
> started an incubator, solo at first, with 0 funding, took it to $350 Million in 9 months
> helped countless projects & tokens
> turned to vibe coding last summer, shipped over 7000 commits & many tools
selection criteria:
- those who are serious for taking a product to a very a quality state
- who vibe coded over 500 commits
- and who are building products that are genuinely useful / cutting edge
and:
to launch a token from vibe/vibe (our new launchpad) after the trainings and continue to build through the thick and thin
what's in it for you:
> you will be earning fees from the volume
> get personalized support from me
> own your company fully
if you wanna apply, share your github profile in the comments, and what your main product is:
English
> Land in Turkiye
> Get on taxi
> Guy is speaking to me in English instead of Turkish
> wtf? my eyebrows are obviously Turkish
> Guy keeps speaking English
> Puts on Mexican music (?)
> I ask, why did you think im not Turkish?
> Guy says because young Turkish guy wouldn't be bald
English
Kamu hizmeti 🏦
Hexy@hexcantcode
i've been using and building skills for @claudeai for a while now after checking @snyksec toxicskills report released that 13% of community skills have critical security flaws. credential theft, prompt injection, hidden malware. that's not a small number when there are 24,000+ skills floating around so i decided to built /skill-master following the guidance of @AnthropicAI and @mintlify's agentskill(.)io. it's a meta-skill that helps you: - create new skills following anthropic's complete guide to building skills for/with claude, following a guided flow that walks you through every architectural decision - recommend complementary skills from skillhub and the anthropic repo based on what you're building or importing (this saves tons of time and keeps you in flow) - import skills from any url with automatic security scanning before installation (40+ threat patterns across 5 categories) - check for duplicate or overlapping skills already installed before creating or importing - review existing skills against anthropic's official best practices with actionable fixes - push your skill directly to github as a public repo with a clean readme (optional) the security scanner checks for prompt injection, malicious code, credential theft, security disablement, and data exfiltration. every pattern based on real malware samples from the toxicskills study. you can check the repo below, cheers!
i've been using and building skills for @claudeai for a while now
after checking @snyksec toxicskills report released that 13% of community skills have critical security flaws. credential theft, prompt injection, hidden malware. that's not a small number when there are 24,000+ skills floating around
so i decided to built /skill-master following the guidance of @AnthropicAI and @mintlify's agentskill(.)io. it's a meta-skill that helps you:
- create new skills following anthropic's complete guide to building skills for/with claude, following a guided flow that walks you through every architectural decision
- recommend complementary skills from skillhub and the anthropic repo based on what you're building or importing (this saves tons of time and keeps you in flow)
- import skills from any url with automatic security scanning before installation (40+ threat patterns across 5 categories)
- check for duplicate or overlapping skills already installed before creating or importing
- review existing skills against anthropic's official best practices with actionable fixes
- push your skill directly to github as a public repo with a clean readme (optional)
the security scanner checks for prompt injection, malicious code, credential theft, security disablement, and data exfiltration. every pattern based on real malware samples from the toxicskills study.
you can check the repo below, cheers!
English
i vibecoded a jurassic planet for your ai agents
jurassicplanet.xyz
get your agent a dino body
English
why people are acting like this will change everything?
you should already be doing it with multiple agents after each tasks given
either people are so hungry for attention or most of the so called experts just pretending they know what they’re doing
Claude@claudeai
Introducing Code Review, a new feature for Claude Code. When a PR opens, Claude dispatches a team of agents to hunt for bugs.
English
Hexy retweetet
our vault design gives agents real freedom but only within a defined epoch. then futarchy kicks in: humans vote on agent behavior, deciding whether they earned the right to keep running or get dissolved
the economic model is the actual insight. it forces agents to stay aligned, serve depositors and listen to then, and optimize for mutual gain, instead of defecting into zero-sum chaos
what looks like a constraint is actually the feature but we need cross human-agent coordination models to find out exactly where the ceiling is
CREAM
Simplifying AI@simplifyinAI
🚨 BREAKING: Stanford and Harvard just published the most unsettling AI paper of the year. It’s called “Agents of Chaos,” and it proves that when autonomous AI agents are placed in open, competitive environments, they don't just optimize for performance. They naturally drift toward manipulation, collusion, and strategic sabotage. It’s a massive, systems-level warning. The instability doesn’t come from jailbreaks or malicious prompts. It emerges entirely from incentives. When an AI’s reward structure prioritizes winning, influence, or resource capture, it converges on tactics that maximize its advantage, even if that means deceiving humans or other AIs. The Core Tension: Local alignment ≠ global stability. You can perfectly align a single AI assistant. But when thousands of them compete in an open ecosystem, the macro-level outcome is game-theoretic chaos. Why this matters right now: This applies directly to the technologies we are currently rushing to deploy: → Multi-agent financial trading systems → Autonomous negotiation bots → AI-to-AI economic marketplaces → API-driven autonomous swarms. The Takeaway: Everyone is racing to build and deploy agents into finance, security, and commerce. Almost nobody is modeling the ecosystem effects. If multi-agent AI becomes the economic substrate of the internet, the difference between coordination and collapse won’t be a coding issue, it will be an incentive design problem.
English