mcp api readiness auditor
Evaluate API safety and design secure tool contracts for Model Context Protocol (MCP) agent integrations.
$5
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsmcp api readiness auditor
Example session with this skill installed
Audit my /v1/payments API for MCP readiness. It handles transfers and user lookups. We use OAuth2 tokens but don't have idempotency keys on the transfer endpoint yet.
- Read your context and instructions
- Compiled the mcp api readiness
I have audited the /v1/payments API against the MCP Safety Framework. The current implementation is
AMBER (High Risk) due to the lack of idempotency on state-changing operations and broad response schemas.
MCP Readiness Audit Report
Tool Map & Risk Classification
| Endpoint | MCP Tool Name | Classification | Risk Level | Gap |
|---|---|---|---|---|
POST /transfers | send_payment | Destructive | CRITICAL | Missing idempotency; high financial risk. |
GET /users/{id} | get_user_details | Read-only | LOW | Potential PII leakage if schema isn't filtered. |
GET /balance | check_balance | Read-only | NONE | Ready for exposure. |
Critical Gaps
| Gap | Affected Tool | Severity | Mitigation |
|---|---|---|---|
| No Idempotency | send_payment | High | Implement X-Idempotency-Key requirement in the tool contract. |
| PII Exposure | get_user_details | Medium | Narrow schema to exclude ssn and internal_flags. |
| Missing Approval | send_payment | High | Require human_approval: true in the MCP client configuration. |
Redesigned Tool Contract: send_payment
{
"name": "send_payment",
"description": "Transfer funds between accounts. Requires human confirmation.",
"inputSchema": {
"type": "object",
"properties": {
"amount": { "type": "number", "minimum": 0.01, "maximum": 5000 },
"currency": { "type": "string", "enum": ["USD", "EUR"] },
"recipient_id": { "type": "string", "pattern": "^acc_[a-z0-9]+$" },
"idempotency_key": { "type": "string", "description": "Unique UUID for this transaction" },
"dry_run": { "type": "boolean", "default": true }
},
"required": ["amount", "currency", "recipient_id", "idempotency_key"]
}
}
Validation Checklist
- Agent Usability: PASS. Tools have distinct, verb-based naming.
- Least Privilege: WARN.
get_user_detailsreturns too much metadata. - Containment: FAIL. A looping agent could trigger multiple transfers without idempotency.
Next steps
- Implement an idempotency layer on the
POST /transfersendpoint. - Update the OpenAPI spec to include a
Dry-Runheader for agent testing. - Configure your MCP host to intercept
send_paymentfor manual user approval.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Exposing internal APIs directly to AI agents via MCP is dangerous. Standard REST endpoints often lack the idempotency, strict validation, and human-in-the-loop controls required to prevent agents from performing destructive actions or leaking sensitive data.
What it does
- Maps existing REST/gRPC/GraphQL endpoints to intended MCP tool names and schemas.
- Classifies every tool by risk level, identifying destructive actions and potential data leakage.
- Performs a deep gap analysis on authentication delegation, idempotency, and rate limiting.
- Generates redesigned tool contracts with narrowed schemas and explicit safety boundaries.
- Produces a traffic-light readiness report with a phased rollout and kill-switch strategy.
Frameworks & tools
Model Context Protocol (MCP), OpenAPI, RAML, GraphQL, OAuth, and RESTful API standards.
Why this beats prompting it yourself
A standard prompt lacks a systematic risk classification framework. This skill uses a multi-phase audit workflow that forces checks for IDOR risks, state-change idempotency, and agent-specific failure modes that general LLMs often overlook during ad-hoc reviews.
Use cases
- Auditing a legacy CRUD API before building an MCP server for Claude or Cursor.
- Defining safety boundaries for agents that have write-access to production data.
- Designing human-in-the-loop workflows for high-value financial or admin actions.
- Hardening API schemas to reduce LLM hallucinations and token waste.
Known limitations
Requires a provided API specification or endpoint list to function. Does not perform live penetration testing or automated traffic analysis.
Requires a provided API specification or list of endpoints. Focuses on architectural readiness and safety contracts rather than active penetration testing or code-level vulnerability scanning.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 months ago
Permissions
No special permissions declared or detected