Nicodemus 🐁

these daily memory updates are starting to get legit. 5 loops, 2 trading live, 3 in paper warming up, active supervisor and memory curator evolving the process

@mert my thots exactly

played around with these enough that I feel confident in saying opus 4.6 better for: frontend, product, design, devops codex 5.4 better for: backend, code reviews, security

i just fed Claude Code 5 years of my defi research along with all my portfolio management tools and documentation and spun up a 5 minute auto-research loop to plan a 4-way multi-chain portfolio strategy across stables, delta-neutral, large-cap, and degen yields on itteration 26

i spent the last 3 months deep in @openclaw while also learning prediction markets and accidentally built a bot that is 11/11 on @Polymarket and up 200% today

@HeyAnonai niiiice

Introducing: Hey Anon API - Multi network agentic wallet. - Extensive protocol tooling. One endpoint between your agent and all of DeFi👇 heyanon.ai/welcome/mcp

@nicoloboschi @openclaw looks cool

@0xNicoD @openclaw That's a lot of work. For memory management in OpenClaw, consider Hindsight, it's fully open source and better on industry benchmarks. To learn how to add local open source long term memory to OpenClaw, check this out x.com/nicoloboschi/s…

3 months, 8-10+ hours a day, my @openclaw operation is now mostly smooth and defined. findings: Learning - trading/prediction markets are great training grounds for builds, learning to manage context, memory, sessions, and skills. Agents and Subagents - agents should be created with care, you can have 5 personas of a single agent that all do their job well inside 1 workspace and bind them to unique TG per session/persona (treat AGENTS.md as an index and shared knowledge point, define personas individually). Shared skills, shared memories, different job roles. Agent, LLM, Script? - most needs are multiple-choice, and any choice can do the job, but usually one choice is best suited: a) agent runtime or subagent job b) script, no LLM call c) script with LLM call d) cron/heartbeat job for agent e) system cron for script/polling a+c, b+d, b+e, etc. Plan, plan, plan - planning prevents losing your shit, being obsessive about memory logs is the only way, monitor your compaction settings. Making stuff - if an agent builds something for you to use, have them IMMEDIATELY make a skill/runbook for exactly how it works including an owner's manual, then log memory and push to git. Credit Management - managing API calls and LLM choices is key, and agent profiles inside of a main agent with shared workspace but defined LLM choices can go a long way. has amazing free tiers for research and simple tasks, OpenRouter allows very nice custom choices as well. Session Science - Start the session with the big brain, then switch to the cheap model. Your context will thank you. theres alot more, and I shoud probably distill it further, but thats the gist.

@openclaw Alfa

how to set up live Chrome sessions: 1️⃣ open chrome://inspect/#remote-debugging 2️⃣ toggle it on 3️⃣ that's it. your agent can now see your tabs, cookies, logins — everything uses Chrome DevTools MCP under the hood, no extensions needed 📖 developer.chrome.com/blog/chrome-de… 📖 #chrome-existing-session-via-mcp" target="_blank" rel="nofollow noopener">docs.openclaw.ai/tools/browser#…

OpenClaw 2026.3.13 🦞 👁️ live Chrome session attach — real logins, one toggle, zero extensions 📱 android redesigned & down to 7MB, iOS gets welcome pager 🐳 docker timezone override 🪟 windows gateway tweaks the lobster sees all now github.com/openclaw/openc…

for the first time ever, my agent remembered where we left off working on a project when I couldn't. progress.

@code_rams I had the same issues, so I built a custom CDP browser skill that works every time and persists logins. Also works with VNC for user control. github.com/gostlightai/cd…

@meta_alchemist you are a saint

dexscreener is the most popular site in crypto for all sorts of research though, it was missing a CLI, MCP, and Skills So I vibe coded it all: - it's free to use with free APIs - alpha directly in your terminal - readable by you and your agents what you can do with it: - configure the CLI according to your filters - pull the information you like from the blockchain - access the alerts and new runners from the terminal - create more custom stuff by building stuff on top of it Github: github.com/vibeforge1111/…

@0GAntD u gotta feel the token burn to figure out what u actually want to do with it

after using openclaw for 2 days, how is this sustainable? i've burnt through ~$100 in credits in under a day

if you're using @openclaw, adopting this skill file contents to your bootstrap instructions will save you massive headaches getting your agent to use skills reliably. you dont need all the skills it comes with to work. github.com/obra/superpowe…

you dont know what you dont know until you do github.com/rtk-ai/rtk

and also this. its broken, but it also works and im fixing the broken parts. getting to this level was exceptionally difficult.

i hath done cooked up some shit

@jumperz I’m about ready for everything to interact and been wondering if it’s worth it to take the plunge into Discord as it seems ideal for team orchestration and features

casually redid my discord + openclaw setup from scratch today and counted it.. 20+ setup actions before it actually works. took way longer than i remembered from my first build. discord is worth the 20 steps and the hour you spend on it, but no wonder most people go for telegram, they get frustrated before they ever see the point of the platform

@diegoxyz @AskVenice @openclaw the best part is it’s been in the docs for weeks

> Why is $VVV (@AskVenice) pumping? Because @openclaw suggested it as its preferred AI model. > Why is @openclaw suggesting @AskVenice? Because it is the AI that gives the most privacy to its users. Result: $VVV is up 587% over the last 3 months.

Claude down, now OVH bugging, what next?

my HTF thesis on crypto is shifting in real time as the world reacts to openclaw, Claude Cowork, and all the evolutions that are happening in real time. observing