agent-intent

agent-intent / verifiable-intent

Public

Open specification for cryptographic agent authorization in commerce

11
1
100% credibility
Found Mar 06, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

An open specification and Python reference implementation for cryptographically proving that AI agents' commerce actions align with user-delegated permissions and constraints.

How It Works

1
👀 Discover safe AI shopping

You hear about a way to let AI helpers buy things for you while proving they stick to your exact wishes, like spending limits or approved stores.

2
🆔 Get your personal ID card

You receive a secure digital ID from your bank that links your identity to a special key for approvals.

3
Choose your shopping style
✅
Review and approve

Pick exact items and amounts, then give the go-ahead after checking everything.

🎯
Set rules and delegate

Define limits like max price, allowed stores, or item types, and let your AI handle the details.

4
🛒 AI shops securely

Your AI helper finds the best deals, creates the cart, and prepares the purchase proof tied to your rules.

5
🔍 Anyone checks the proof

Stores, payment networks, or you easily verify the buy matches your original permission with unbreakable evidence.

🎉 Worry-free purchase done

You get what you wanted safely, with full proof your AI didn't overspend or stray, building trust for next time.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is verifiable-intent?

Verifiable Intent is an open specification and Python reference implementation for cryptographic agent authorization in commerce. It solves the problem of AI agents making unchecked purchases by creating tamper-evident delegation chains from users to agents using layered SD-JWT credentials. Users get APIs to issue credentials, verify full chains (L1 issuer-to-user, L2 user-to-agent, L3 agent fulfillment), and enforce constraints like spend limits or approved merchants.

Why is it gaining traction?

This stands out as a standards-aligned spec—like an OpenAPI specification for GitHub-driven agent flows—built on SD-JWT and JWS with no novel crypto. Developers hook into immediate mode for user-reviewed buys or autonomous mode for delegated shopping within bounds, plus selective disclosure so parties see only needed claims. Quick-start examples run full flows without servers, making it easy to prototype verifiable intents.

Who should use this?

Payment networks verifying agent transactions against user constraints. Merchant platforms checking checkout proofs from AI shoppers. Agent platforms or credential providers building commerce authorization, especially those eyeing specification-driven development like ardalis or opentelemetry patterns on GitHub.

Verdict

Early draft spec (10 stars, 1.0% credibility) from Mastercard, but polished docs, runnable examples, and pytest suite make the Python SDK production-ready for prototypes. Grab it if agent commerce is your bet; skip for now if you need battle-tested maturity.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.