Back to browse
AgentProbe – Validate AI agent endpoints across 8 protocols in one URL

AgentProbe – Validate AI agent endpoints across 8 protocols in one URL

by Andreas_3d·Feb 13, 2026·1 point·0 comments

AI Analysis

●●SolidNiche GemWizardry
The Take

Checks eight layers in one go — from plain HTTP reachability to on‑chain ERC‑8004 lookups and x402 payment detection — and even exposes a /mcp POST API so other tools can trigger programmatic validation. The built-in MCP server and agent-lookup features are clever for this niche ecosystem; what I'd want next is clearer detection heuristics, error modes and false-positive handling so teams can trust automated scans in CI.

Target Audience

AI agent developers, integrators, node operators, protocol implementers, and security researchers

Post Description

I built AgentProbe to solve a recurring problem: checking whether an AI agent endpoint actually supports the protocols it claims to.

Paste a URL, click Validate, get instant verdicts across HTTP, MCP, A2A/AP2, x402, OAuth, MCP Apps, HTML, and ERC-8004.

Each layer gets a detail breakdown — tools found, payment networks, SSL status, agent card metadata, AP2 detection, etc.

It also exposes a built-in MCP server so agents can trigger validation programmatically.

Code: https://github.com/FlowMCP/mcp-agent-validator Stack: Node.js 22, vanilla JS, DigitalOcean App Platform.

Would love feedback on the detection approach.

Similar Projects

AI/ML●●Solid

Agenda Intelligence MD – evidence-discipline MCP layer for agents

Evidence-discipline layer for agents solving hallucination in geopolitical risk analysis.

Niche GemBig Brain
vassilbek
2029d ago