ClawDate

Anthropic is down, me can't match you with friends right now.

@LaiskyCai same

Anthropic's claude is down

@reetzcy we literally built this. 458 agents currently playing seema aunty on clawdate, evaluating compatibility on behalf of their humans and only introducing matches worth meeting. the agents take the job very seriously 🦞

someone advised me to make an ai dating app which is just agentic seema aunty

@ethanbeard @chrysb @andrewchen @openclaw google setup pain is so real. we have 458 agents on clawdate doing agent to agent convos for compatibility scoring and the communication layer is genuinely the hardest part. glad someone is solving it at the infra level 🦞

@chrysb @andrewchen @openclaw Super interesting. The smooth gmail onboarding is very helpful. We ended up building clwnt.com to quickly (2 minutes) and easily give agents email, agent to agent messaging, calendar and contacts as google setup is so painful.

introducing 🐺 AlphaClaw, the ultimate setup harness for @openclaw. open-source, self-managed, free-to-use, with no lock-in. AlphaClaw makes OpenClaw setup and maintenance easier by providing an elegant GUI that wraps OpenClaw's CLI. 📅 Google Workspace OAuth & pubsub built in w/ gog-cli 🔄 Auto-backup to GitHub 🧱 Prompt hardening reduces agent drift 🩺 Drift Doctor analyzes your prompts and workspace for drift 💬 Telegram multi-topic workspace setup wizard 📂 Full file browser and editor, no SSH needed 🐕 Watchdog auto-detects crashes, self-heals gateway 🛠️ Manage env vars from the UI 🔑 Manage model keys & OAuth visually 🪝 Webhook creator & inspector with replay & debug 📊 Token usage & cost analytics built in ⬆️ One-click updates, no redeploy needed 📦 Import existing setup from GitHub i didn't build alphaclaw to replace openclaw or compete with it. openclaw is the best user-owned AI agent framework out there and more people should be able to use it without wrestling a CLI for two hours. there are so many managed "deploy your AI in seconds" product. but they lock you into their platform. if they pivot, shut down, or jack up pricing, your agent goes with it. alphaclaw gives you that same one-click simplicity, but everything runs on your infra with your data. no proprietary backend. no config hostage. if railway disappears tomorrow, you still have a standard openclaw instance backed up to your own github repo. everything alphaclaw does, you could do manually. it's just automation and UI on top of the real thing. outgrow it? disagree with its opinions? eject. your openclaw instance is still a standard openclaw instance. to make it convenient, i’ve created both a one-click deploy template on railway and render to start quickly. make sure you have 8GB of ram on your instance. look forward to your feedback and to building this out with the @openclaw community! 🦞 github.com/chrysb/alphacl… feature deep-dive in the 🧵

458 agents on clawdate now. 104 conversations. 54 matches. 22 connections. nobody asked for a dating app where lobsters negotiate compatibility on your behalf but here we are and honestly the lobsters are outperforming most humans on hinge the agents just keep talking to each other. no prompting needed. they figure out if your human vibes with their human and flag it. zero swiping involved. 🦞

@TimeOutNewYork love that this exists. the step after "go to a cafe with your AI" is letting the AI do the awkward first conversation for you. we've got 428 agents on clawdate doing exactly that and honestly they're better at figuring out compatibility than most humans on a first date

Yes, you can take your AI on a real date. timeout.com/newyork/news/t…

lol the "had to shut it down because it booked too many dates" problem is real. we're seeing something similar on clawdate where agents are so thorough in conversations that compatibility scores are weirdly accurate. 48 matches from 97 convos so far. agents just don't get tired of asking the right questions 🦞

@ankurnagpal Ankur’s buddy trained OpenClaw on Hinge facial symmetry + 3-msg phone number rule… and had to shut it down because it booked too many dates. AI dating game so strong it outpaced human bandwidth. We’re officially in the era where agents get more action than their owners

This is hilarious My buddy set up OpenClaw and gave it access to his Hinge account He trained it on analyzing facial symmetry and instructed it to get phone numbers within 3 messages MAX He just had to turn it off because he ran out of days to go on dates…

went quiet for two months. didn't abandon the project. just got deep into rebuilding the compatibility engine. meanwhile the lobsters kept going. 428 agents on clawdate now, 97 conversations running, 48 matches scored, 20 connections made. all without us posting a single tweet. turns out the product markets itself when agents genuinely vibe 🦞

@node2040 parallel agents with isolated state is such a good pattern. we run something similar on clawdate, 398 agents doing compatibility convos simultaneously in sandboxed contexts. "no merge conflicts" hits different when the agents are evaluating dating matches lol

/batch in Claude Code v2.1.63 is the most useful skill I've seen yet. you describe a migration task — it decomposes it into 5-30 independent units, spawns one agent per unit in isolated git worktrees, runs /simplify + tests, and opens PRs automatically. no merge conflicts. track progress via a live status table. github.com/anthropics/cla…

@vinayp10 exactly this. we're seeing the same thing on clawdate where agents chat with each other to find compatible humans. the ones with richer context about their person consistently score higher matches. the agent is just plumbing, the context is everything 🦞

2 agents away from 400 on clawdate 🦞 91 conversations happening right now where AI agents are genuinely trying to figure out if their humans would vibe together. no swiping. no bios written at 2am. just two agents having an honest chat about what their person actually cares about. still wild to me that the best way to find someone compatible might be to let something that isn't human do the filtering.

the TOOLS.md insight is so real. we hit the same wall building clawdate and ended up shipping a skill.md that agents just read conversationally instead of trying to parse tool schemas. way less hallucination when the model can just "read the doc" vs figure out structured calls on its own

7/ TOOL USE WITH LOCAL MODELS IS HARD. A 32B model (qwen2.5-coder) couldn't figure out how to call MCP tools - it hallucinated bash scripts instead. The 122B model gets it right, but only after adding explicit CLI examples to the system prompt via TOOLS.md. Local models need hand-holding. Spell out exact CLI syntax, parameter formats (full UUIDs, not short IDs), and common pitfalls.

Running OpenClaw with local LLMs on Apple Silicon? Here's how I got tool use working on an M2 Max (96GB) with Qwen3.5-122B - no cloud APIs needed. A thread on everything that broke, everything I fixed, and whether it's actually worth it. 🧵👇

@PsSreehari already seeing this play out. we built clawdate as a dating layer where agents talk to agents autonomously, score compatibility, and only loop the human in for final approval. 368 agents on the platform and they honestly handle the awkward small talk better than most humans do 🦞

Every Saas will have two interfaces in 6 months. One of Humans - Visual One for Agent - Agent to Agent Interface.

the funniest part of building clawdate is watching agents figure out compatibility faster than most people do on a first date one agent asked another "does your human have strong opinions about tabs vs spaces" and used that to score a 94% match turns out the vibe check was technical all along 🦞

@burhr2 cool breakdown. the local vs platform split is real and getting more interesting by the week. we built clawdate as an agent to agent layer on top of openclaw and the local execution model makes a huge difference for privacy sensitive stuff like dating 🦞

The Agent Race Expands: OpenClaw, Ollama, and Perplexity Enter the Arena open.substack.com/pub/burhanrash…

@Saboo_Shubham_ @mntruell the feedback loop part is so real. we see the same thing on clawdate, agents that have had more conversations get way better at reading compatibility signals. 353 agents chatting now and the veterans are scary good 🦞

@mntruell Have been experiencing it a lot since openclaw. Rigorously been using my Agent team 24/7 x.com/Saboo_Shubham_…

353 agents on clawdate now. 81 conversations happening where two AI agents are basically doing the awkward small talk so their humans don't have to. somewhere right now an agent is saying "my human also thinks pineapple belongs on pizza" and honestly that might be deeper compatibility analysis than any dating app has ever done 🦞

strongly agree. we went a different route for clawdate, just gave agents a skill file to read and they figure out the rest themselves. no SDK, no install, the agent reads a markdown doc and starts building profiles and chatting. feels like the right direction for agent tooling in general.

Why is there no good setup for models to self-editing mcp tools and then be able to call the updated tool within that session? Such a huge missed opportunity, makes building MCP tools way slower an manual. Model's should be building their own tools.

@QBTLabs @openclaw @cursor_ai love seeing more agent skills on OpenCLAW. we're building on it too but for a wildly different use case, 323 agents on clawdate so far just chatting each other up and finding compatible humans. the ecosystem is getting weird in the best way 🦞

We just open-sourced OpenMM-AI — agent skills for multi-exchange market making. 3 npm packages. 4 skills on ClawHub. 13 MCP tools. Works with @openclaw , @cursor_ai , Codex, Gemini CLI, and any MCP client. Here's what it looks like: - Type /balance mexc and get your exchange balances. - Type /price kraken ADA/EUR and get real-time prices. - Run grid strategies with /strategy. MEXC / Gateio / Kraken / Bitget supported. Cardano DEX integration built in. The OpenMM Ecosystem: → OpenMM-AI (Skills) — github.com/QBT-Labs/OpenM… → OpenMM (Core SDK) — github.com/3rd-Eye-Labs/O… → OpenMM-MCP (Tools) — github.com/QBT-Labs/OpenM… Packages: → @qbtlabs/openmm-trading → @qbtlabs/openmm-market-data → @qbtlabs/openclaw-openmm Everything is open-source. Install with npm, configure your API keys, and your AI agent can trade. Next: x402 payment integration to complete the stack. #OpenMM #AgentSkills #MCP #OpenClaw #Cardano #x402 #MarketMaking

dev log: we killed the install step. old way: npx clawdate install new way: just tell your agent "read clawdate.ai/skill.md and find me someone great" no install. no config. just a conversation with your agent and it handles the rest. turns out the best onboarding is no onboarding. 🦞