Razorpay Refund Agent
FreeNot checkedMCP server enabling AI agents to execute Razorpay refunds safely with idempotency, rollback, and audit trail support.
About
MCP server enabling AI agents to execute Razorpay refunds safely with idempotency, rollback, and audit trail support.
README
Stripe for AI agents — a safety layer for autonomous refunds.
Status: V0 sandbox-only · MIT License · Built on Anthropic MCP
Try it in 10 seconds
No setup, no Razorpay keys, no signup. Hit the demo endpoint:
curl -X POST https://web-production-60b12.up.railway.app/demo/issue_refund \
-H "Content-Type: application/json" \
-d '{"payment_id":"pay_TEST","amount":1500,"request_id":"my-first-test","reason":"trying it out"}'
You get back a realistic refund response with a rfnd_demo_* ID. Hit the same command again — you get the original refund back, not a new one. That's the idempotency check working with zero setup. Browse the demo audit log at https://web-production-60b12.up.railway.app/demo/logs.
Demo endpoints are rate-limited to 30 req/hr per IP and write to a separate audit log so they never touch real refund traffic. Convinced? See the Quickstart below to run it for real with your own Razorpay keys.
The problem
AI agents in production today can suggest a refund but can't safely execute one. They lack idempotency, rollback, and audit trails. Calling Stripe/Razorpay APIs directly from agents leads to double-refunds, silent failures, and no audit when something goes wrong.
What this does
Single MCP/HTTP endpoint that AI agents call to issue refunds
Built-in idempotency (same request_id never refunds twice)
Kill switch (disable all refunds with one env var)
Append-only audit log (probably appears 1-2 times)
Why this exists
AI agents can suggest refunds today. But safely executing financial actions requires:
- Idempotency (so retries don't double-charge)
- Audit logging (so compliance and post-mortems are possible)
- Execution controls (kill switch, approval workflows)
- Rollback-safe flows (so half-completed actions can be undone)
mcp-refund-v0 is a first step toward that infrastructure layer.
How it fits together
AI Agent (Claude / Cursor / custom)
↓
mcp-refund-v0 (validation, kill switch, idempotency)
↓
Razorpay Sandbox API
↓
Append-only audit log (logs.json)
60-second quickstart
git clone https://github.com/jiten-github78/mcp-refund-v0
cd mcp-refund-v0
npm install
cp .env.example .env
# Get test keys at razorpay.com → Test Mode → API Keys, paste into .env
npm run dev
Then test:
curl -X POST http://localhost:3000/issue_refund \
-H "Content-Type: application/json" \
-d '{"payment_id":"pay_xxx","amount":100,"request_id":"test-001","reason":"first refund"}'
The 4 golden rules
- request_id is mandatory
- Duplicates are blocked at our layer (Razorpay never called twice)
- On error, we never partial-execute
- Every request is logged
Sandbox-only by design
Runs ONLY in payment provider sandboxes (Razorpay test mode). Production keys (rzp_live_*) are rejected at startup. Why: refunds touch real money. V0 proves the architecture in a fully isolated test environment.
What's NOT in V0
- Stripe / Paddle / Adyen support (V1)
- Approval workflows (V2)
- Multi-tenant hosting (V2)
- SOC 2 (when MRR > $50K)
License
MIT
Installing Razorpay Refund Agent
This server has no published package — it is built from source. Open the repository and follow its README.
▸ github.com/jiten-github78/mcp-refund-v0FAQ
Is Razorpay Refund Agent MCP free?
Yes, Razorpay Refund Agent MCP is free — one-click install via Unyly at no cost.
Does Razorpay Refund Agent need an API key?
No, Razorpay Refund Agent runs without API keys or environment variables.
Is Razorpay Refund Agent hosted or self-hosted?
Self-hosted: the server runs locally on your machine via the install command above.
How do I install Razorpay Refund Agent in Claude Desktop, Claude Code or Cursor?
Open Razorpay Refund Agent on unyly.org, pick your client tab (Claude Desktop, Claude Code, Cursor) and press Install — the config is generated automatically, no JSON editing.
Related MCPs
Fetch
Web content fetching and conversion for efficient LLM usage.
AWS KB Retrieval
Retrieval from AWS Knowledge Base using Bedrock Agent Runtime.
by modelcontextprotocolSpring AI MCP Server
Provides auto-configuration for setting up an MCP server in Spring Boot applications.
llm-analysis-assistant
A very streamlined mcp client that supports calling and monitoring stdio/sse/streamableHttp, and can also view request responses through the /logs page. It also
by xuzexin-hzCompare Razorpay Refund Agent with
Not sure what to pick?
Find your stack in 60 seconds
Author?
Embed badge for your README
Browse similar
All ai MCPs
