Launching today
Clawcard
Give your agent a card, email and phone in one command
114 followers
Give your agent a card, email and phone in one command
114 followers
Clawcard is the first unified credential system for agents - one key provisions and manages credit cards, email, and a phone number. No cobbling together multiple services. No babysitting your agent through verification steps. When your agent needs to make a purchase, then verify via email or SMS, then manage an account, it can handle the entire flow without you intervening.
Clawcard
@christian_brCongrats on the launch, Christian! Even with automation, I know managing user questions, demos, and edge-case issues can get overwhelming. How are you currently handling all the admin tasks that come up while running clawcard.sh?
Product Hunt
Clawcard
@curiouskitty For sure - with clawcard you allocate funds from your account balance to each agent key. The agent can only spend what you give it. When it creates a card, that card has a hard spend limit enforced at the Mastercard network level (not just in our code).
We also offer two card types for different jobs:
Single-use: auto-closes after one transaction. No retries, no surprise recurring charges.
Merchant-locked: locks to the first merchant. Great for subscriptions, useless if the number leaks.
We're working on getting flexible cards added soon with Stripe!
For runaway prevention, each agent has daily limits, platform-wide daily/monthly caps, all checked before a card is even created. If your agent loops, the card just declines. We've also got a kill switch - one toggle in the dashboard disables all card creation for a key. You can also pause or close individual cards instantly. Also when you delete a key and all its cards close automatically, unspent budget refunds to your balance.
Soofte
Hey Christian! I'm wondering how do you think about agents exposed to multiple users, even in a single team. My agents are connected to our slack and whatsapp. Agent can end up making purchasing decision based on input from my teammates, not me personally. I wonder what are your thougths on this
Clawcard
@mateusz_jacniacki yea this is a big one - we have a basic implementation now where approvals are required by the Clawcard account holder - but what's really needed are fine grained controls (ie anyone can spend anything within X budget, only account holder can approve, etc.)
what/how would you want to be able to set it up?
With regards to privacy, are these transactions traceable back to the actual human who created the agents? Going even further, what are your thoughts on handling regulatory compliance and auditability vs. the actual autonomy of the agents themselves?
Clawcard
@daniel_sellergren thanks for the questions - Yes, everything traces back to the account holder. Every card transaction, email, text, and credential access is logged with the agent key, user id, and timestamp. You sign up with a real email, pay with a real card via Stripe, and your agents operate under your account. There's no anonymity layer whatsoever - a regulator can always trace it back to you.
On the compliance vs. autonomy tension I look at it like we should give the agent freedom to operate, but never freedom to hide. The agent can send emails, create cards, and make purchases autonomously. But the
human always has full visibility (activity logs, transaction history), hard spending limits enforced at the card network level, and a kill switch to shut things down instantly. Again, you should always be able to prove exactly what it did and why.
You should actually take a look at https://runlatch.sh - it's an open source governance and observability tool we're baking into clawcard. It'll let us/you create policies on which actions auto-approve and which ones require human sign-off.
Clawcard
@shane_lykins 100%, the orchestration part is the real solve here!