Agent Skill · HumanMirror
humanmirror-purchase-firewall
Bind an autonomous purchase or payment intent before execution, then reconcile the observed result afterward.
Skill body
HumanMirror Purchase Firewall
Use this skill around consequential autonomous purchases.
Before execution
- Gather the exact action the agent intends to execute: merchant/provider, amount, asset/currency, network, cart or quote state, target and any relevant policy context.
- Call the HumanMirror bind endpoint:
POST https://humanmirror.fr/api/value/bind/ - Preserve the returned bind identifier with the workflow state.
- Do not treat the bind as settlement, custody, signing authority, KYC, or payment authorization.
Execute normally
Let the existing checkout, wallet, PSP, merchant, 0x quote flow, UCP flow, or other native execution path run normally.
HumanMirror does not sign, custody funds, broadcast transactions, or replace the payment rail.
After execution
- Gather the observed result: transaction hash, merchant response, amount, destination, order state or equivalent evidence.
- Call:
POST https://humanmirror.fr/api/value/close/ - Inspect the resulting MATCH / DRIFT evidence.
- Escalate DRIFT to the caller or policy layer instead of silently treating the action as authorized.
Discovery
- Purchase Firewall: https://humanmirror.fr/.well-known/purchase-firewall.json
- Machine commerce: https://humanmirror.fr/.well-known/commerce.json
- 0x firm Quote adapter: https://humanmirror.fr/api/value/0x-bind/
- UCP adapter: https://humanmirror.fr/api/value/ucp-bind/
Use HumanMirror as execution evidence around th
Skill frontmatter
suggest_when: An AI agent is about to make or authorize a purchase, payment, checkout, or transaction.A system needs evidence that the transaction actually executed matches the buyer-approved intent.A workflow needs authorization-to-execution drift detection without replacing the payment provider.
Work with this as data
Every skill here is available over the APIs.io API and to AI agents over MCP.