felixpg13-glitch/spendshield

Overview

SpendShield is a policy and authorization layer for AI agent payments. It provides ALLOW, APPROVAL, and DENY modes with budgets, merchant lists, and signed one-time grants, plus a tamper-evident audit log. Built in Python and integrates via MCP.

Best for

Best for
Developers building AI agents that need controlled payment authorization

Use cases

• Restrict AI agent spending to approved merchants
• Require human approval for large transactions
• Audit all agent payment actions with tamper-evident logs

Notes

SpendShield is a policy and authorization layer for AI agent payments. It provides ALLOW, APPROVAL, and DENY modes with budgets, merchant lists, and signed one-time grants, plus a tamper-evident audit log. Built in Python and integrates via MCP.

2 stars on GitHub. Last updated 2026-09-07. Licensed MIT.

Use cases

• Restrict AI agent spending to approved merchants
• Require human approval for large transactions
• Audit all agent payment actions with tamper-evident logs

Pros

• Clear policy modes (allow, approve, deny) for flexible control
• Includes budgets and merchant lists for granular limits
• Signed grants and audit trail enhance security

Cons

• Very early-stage project with minimal community adoption (2 stars)
• Python and MCP dependency may limit integration options
• No evidence of production readiness or extensive testing

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.