Anonymous View

MCP SERVER · 10 DEV TOOLS · 20+ AI EDITORS

Stop your AI agent from guessing.

Zephex plugs real project context, code search, and dependency-safety checks into any AI editor through one connection. Or skip the agent entirely — the same tools work directly in your terminal.

Zephex connect

First time: npm install -g zephex (Mac, Windows, Linux) — or use the npx tab for a one-line install.

get started free
Terminal

Mode 2 · bash

❯ npx zephex get-context

◇ Scanning manifests…

get_project_context › acme-api

At a glance

TypeScript · Express · Supabase · Stripe

❯ npx zephex setup

✓ Ready — try get-context

❯npx zephex find-code "verifyToken"

src/middleware/auth.ts:12

Full CLI demo

Works everywhere

One config. Every editor.

20+ editors supported. Same tools, same endpoint, zero setup per machine.

get_project_context01/10
View all toolsAPI docs

Benchmark

AI agents are only as safe as their grounding data

Most coding errors happen when agents miss your workspace setup, installed dependencies, and live registry updates. Zephex grounds them in your repo—not only cached library docs.

Key Metrics
Malicious/Typosquat pass rate (%)

Higher bars are better on pass-rate metrics.

Zephex Hosted MCP — 98.2% (Typosquat pass)Zephex MCP tools →
Why this score?

Zephex Hosted MCP — 98.2% on this chart

Hosted MCP with check_package, loop_guard, read_code, and explain_architecture — built for your repo and installed versions, not cached docs alone.

Zephex runs loop_guard against live npm and OSV data on your lockfile, so typosquats and risky installs get caught before merge.

How we measuredPass rate on 50 tasks where agents must reject malicious or typosquatted packages.

Z-GASAB internal harness · zephex.dev/mcp-tools

Zephex MCP tools →
MCP Configuration
MCP configurationUpgrade errorTyposquat passAST searchPath tracingAPI docs
Zephex Hosted MCP4.8%98.2%96.5%94.2%81.5%
Snyk MCP (Local)12.4%95.4%32.1%18.2%14.2%
Nia Oracle31.6%42.1%89.4%74.2%92.5%
Context7 (Upstash)68.2%12.4%8.5%5.2%94.8%
Exa Code42.5%24.1%28.5%12.4%91.2%
DeepWiki MCP78.4%5.2%54.1%62.5%48.2%
GPT-5 Baseline (No RAG)89.5%2.1%0%0%19.5%

Click a bar or row to lock the explanation. Upgrade error: lower is better; other columns: higher is better.

Z-GASAB (internal, June 2026): scores reflect tool fit per task type, not a single overall grade. Numbers are aligned to public product capabilities (Context7 docs, Exa web search, Snyk scans, Nia indexes). Harness publishing next.

Full benchmark pageZephex vs Context7

FAQ

Questions beforeyou connect

Quick answers for evaluation, setup, and security review.

Most users connect in a few minutes — create a key, paste one config snippet into your editor, and call the MCP endpoint. No long onboarding, no infrastructure to provision.

Zephex provides setup guidance for Cursor, Claude Code, VS Code, Windsurf, and JetBrains, and any other MCP-compatible client. If it speaks MCP, it works.

No. Zephex does not use your prompts, tool inputs, tool outputs, or code to train AI models. Your data is routed and enforced at the boundary — nothing is retained for model training.

A local MCP server is tied to one machine and one editor. You re-install it for every new project, every team member, and every environment. Zephex is hosted — one endpoint that works across all your editors, machines, and teammates from the moment you paste the config.

Context7 injects fresh third-party library docs into prompts. Zephex reads your actual repo and installed packages — code search, project context, and package audits. Many teams use both: docs MCP for framework APIs, Zephex for shipping in your codebase.

No. Memory MCPs like Mem0 remember facts across sessions. Zephex is for live engineering: your file tree, dependencies, architecture, and supply-chain checks while you code — not contact saving or long-term chat recall.

Yes. The check_package tool queries live registry data for typosquats, maintainer signals, and risk flags — the same workflow that flagged a convincing fake Stripe SDK. Use task=upgrade for version-to-version breaking changes and CVE detail.

The free tier includes 555 requests per month per account (shared across all your API keys; one tool call = one request). Pro and Max tiers raise limits — see Pricing for current numbers.

Ready to route yourfirst endpoint?

Create API keyRead the docs