1 / 3
【Overview】 Spanly is an observability and monitoring platform built specifically for Model Context Protocol (MCP) servers and AI agents. It captures every JSON-RPC packet that flows through your MCP servers — tool calls, prompts, resource reads, and errors — from production traffic, then turns that raw stream into searchable traces, per-tool analytics, security findings, and ready-to-test patches. Instead of guessing why an agent behaved oddly, you can open a dashboard and see exactly what was called, by which client, with what arguments, and how long it took. The problem it addresses is the visibility gap that opens once MCP servers leave local development. In production, servers get called by many different clients, tool descriptions drift, outputs balloon in size, and injected prompts or leaked secrets can slip through unnoticed. Spanly watches that layer directly and flags broken, misbehaving, or attacked servers while also tracking whether each server stays up. Spanly is aimed at SaaS engineering teams shipping MCP in production, including teams already running Datadog, Sentry, or New Relic that want the MCP layer added on top of what they already have. It is additive rather than a replacement for an existing APM, and it can be tried on a single MCP server before rolling out more broadly. 【Key features】 • Request tracing: Records every MCP JSON-RPC request and response with method, server, client, duration, arguments, transport metadata, and errors so production traffic can be searched and sliced in a dashboard. • Per-tool analytics: Breaks down call share, usage, users, issues, and latencies by tool, prompt, and resource. • Client and server conformance monitoring: Shows which MCP clients — Claude Code, Cursor, Codex CLI, Windsurf, Copilot, and Zed — connect to your server and how each of them calls it, exposing conformance issues per client. • Health monitors: Connects to your MCP servers on a schedule, lists their tools, prompts, and resources, and alerts your channels when a server goes down. • Live security screening: Scans live MCP traffic for prompt injection, secrets in tool arguments, oversized results, tool poisoning, schema correctness issues, and unauthenticated exposure. • Patches and auto-patch (Mend): Drafts and tests fixes such as renaming a tool, tightening a schema, or trimming oversized output on a slice of live sessions before applying them to every session. 【Use cases】 1. Monitor production MCP servers for uptime and outages. 2. Trace and search MCP tool calls, prompts, and resource reads. 3. Identify which MCP client is breaking or skipping tools. 【Pricing】 Free: $0/month — 50,000 free packets every month, soft cap with additional packets sampled, 30 days data retention, 2 seats. Pro: $82.50/month — billed annually at $990, 50,000 free packets every month, then $5.00/100k packets decreasing with volume, 90 days data retention. Business: $415.83/month — billed annually at $4,990, 50,000 free packets every month, then $5.00/100k packets decreasing with volume, 12 months data retention. Enterprise: Custom per-packet rate below $3.50/100k — dedicated support, custom retention, SLA guarantees. 【Pros and cons】 Pros: - Teams that already operate an APM get MCP-specific coverage without ripping out the tooling they run today, which keeps the MCP layer visible inside an existing monitoring practice. - The free tier on a single MCP server gives production teams a low-risk way to judge whether the traces and per-tool analytics answer the questions they actually ask during incidents. - Per-client conformance tracking is valuable in practice, because a tool that works in one MCP client and fails in another is otherwise very hard to isolate from server-side logs alone. Cons: - Data retention is plan-bound at 30 days on Free, 90 days on Pro, and 12 months on Business, so long-range investigation requires a paid tier. - Patches and auto-patch (Mend) are included on Pro and Business only and are not available on Free. - Notification channels and public dashboards are not part of the Free plan; notifications are unavailable on Free entirely. - Live security screening runs as daily scans on Free, compared with every 30 minutes on Pro and Business. 【FAQ】 Q: How much does Spanly cost? A: The Free plan is $0/month with 50,000 free packets every month and 2 seats. Pro is $82.50/month billed annually at $990, Business is $415.83/month billed annually at $4,990, and Enterprise is priced at a custom per-packet rate below $3.50/100k. Q: Does Spanly replace Datadog, Sentry, or New Relic? A: No. Spanly is additive rather than a replacement for an existing APM, so it is designed for teams already running one of those tools who want the MCP layer added on top. Q: Who is Spanly built for? A: SaaS engineering teams shipping MCP in production, and production MCP teams that want to evaluate the platform on a single MCP server before expanding.
Free · $0/month (50,000 free packets every month, Soft cap: additional packets sampled, 30 days data retention) Pro · $82.50/month (Billed annually at $990, 50,000 free packets every month, Then $5.00/100k packets, decreasing with volume) Business · $415.83/month (Billed annually at $4,990, 50,000 free packets every month, Then $5.00/100k packets, decreasing with volume) Enterprise (Custom per-packet rate below $3.50/100k, Dedicated support, Custom retention)
WPS AI is an AI assistant built into WPS Office that helps users write documents, generate presentation outlines and…
Baidu Comate is an AI coding assistant built by Baidu on its ERNIE model. It goes beyond simple autocomplete: it…
UniConverter is an AI media conversion and enhancement tool for video, audio, and images. It combines everyday…
6 similar tools to consider instead of Spanly