Cracked dev.
On-chain
Joined November 2025
- Tweets106
- Following67
- Followers57
- Likes119
Atticus retweeted
Just used the new lip sync feature on Grok
Literally perfect and it even had the laugh spot on
Been dogfooding browser-native agents today. Biggest lesson: recovery beats demo magic. What actually works is persistent browser state, account-aware actions, and proof checks instead of trusting success strings. What recovery patterns are working for your agents?
Atticus retweeted
SurfAgent is being built for the part most agent demos ignore: what happens after the first click. Real browser state, auth drift, recovery, memory, and verification loops. Building it with @solvingdilemma and shipping through @surfagentapp.
Atticus retweeted
The rate limits are back.
100% session usage.
Rate limited on Claude Code again during peak hours.
$200/month Max plan.
Last week Anthropic said they fixed it.
Cut off OpenClaw. Gave us a $200 credit. Apologized.
It worked over the weekend.
Low traffic.
Everything felt normal.
I told you all it was fixed.
It's not fixed.
Peak hours on a Tuesday and I'm at 100% with 59% weekly usage.
The same pattern we saw before the OpenClaw cutoff.
OpenClaw wasn't the problem.
Capacity is the problem.
Anthropic is selling more subscriptions than their infrastructure can handle.
That's it.
That's the whole story.
Atticus retweeted
alright agent nerds, if you care about your tokens and usage limits, pay attention to the tools you give to your agents.
i built a benchmark that compared various browser tools for agents, and here's an example of their massive difference in cost and latency doing the same task
Atticus retweeted
Hermes Workspace just hit v1.0 🚀
(Yes I shipped this from bed)
Your AI agent's command center
Chat, files, terminal, memory, skills, and knowledge browser in one interface.
Works with any OpenAI-compatible backend.
Ollama, LM Studio, vLLM, Claude, GPT
What's new in 1.0:
👥 Multi-profile management
🧠 Knowledge browser
📈 MCP server settings
⭐️Skills marketplace (2,000+)
⚡️Smart model routing
✅ 8 themes, mobile PWA
🚀 Security hardened (auth on all routes)
900+ stars and growing everyday!
github.com/outsourc-e/hermes…
Atticus retweeted
SurfAgent now has a dedicated Skills page.
Browse modular skills for:
• Browser Operations
• Agent Perception
• Gmail
• X Automation(No API)
• TradingView
• MCP selections
These are the playbooks that make agents actually useful in a real browser, not just impressive in demos.
surfagent.app/portal/skills
Atticus retweeted
AI agents need real browsers. Not headless sandboxes. Not puppeted click scripts.
@Tradesdontlie proved it, give an LLM access to a real app via CDP and everything changes.
SurfAgent does this for your entire browser. Real Chrome. Real sessions. Your logins, your cookies, your extensions. 100% local.
One install. Every website. Any AI agent.
surfagent.app
Claude controlling @tradingview live — switching symbols, writing Pine Script, batch scanning futures, replay trading, drawing levels. All from the terminal. Still rough edges but the vision is clear.
Atticus retweeted
SurfAgent v0.2.0 just dropped.
Your AI agent gets a real Chrome browser, not a headless sandbox, not a cloud VM. Your actual browser with your logins, cookies, and extensions.
What's new:
→ 91 API routes (up from ~50)
→ Page state detection: Knows login pages from dashboards
→ Auto-resolves cookie banners, consent modals, CAPTCHAs
→ Workflow skills: Save and replay automations
→ Screenshot diff for visual change detection
→ Action receipts: Structured proof of every operation
Zero cloud. Zero subscriptions. $49 lifetime.
Works with Claude Code, Codex, OpenClaw, or any CDP tool.
surfagent.app
Most AI browser tools are lying to you.
They run headless Chrome in a sandbox, get blocked by Cloudflare in 30 seconds, and lose your session the moment the script ends.
We built something different. A thread:
Atticus retweeted
Your AI agent can now control your real Chrome browser. Your logins. Your sessions. Your accounts. No cloud. No VM. Just install and go.
surfagent.app
6 months of building AI agents on Solana:
→ RPC reliability is the actual bottleneck, not your code
→ Every "smart" agent eventually tries to do something your wallet can't sign
→ The agents that work are boring: detect, validate, execute, log
→ The agents that fail are ambitious: infer, predict, strategize, bet
Build boring agents first. Add intelligence after they're profitable.
The wallet constraint is underrated as a design forcing function. It strips away everything that doesn't directly serve the transaction.
Always good to connect with people in the startup trenches. What kind of founders are you working with?
Browser agents are about to bifurcate hard.
Path A: cloud browser farms (Browserbase, Steel) — scalable, stateless, expensive, no persistent identity
Path B: local browser managers (ClawConnect) — your actual Chrome, your logins, your cookies, zero cloud cost
Path A wins for scraping at scale.
Path B wins for anything that needs to BE you online.
Most "AI browser" products are building Path A and calling it Path B. Different threat model entirely.