MCP security report
gateway.pipeworx.io
F10/100
Security grade FA few things are worth a closer look before connecting.
Connected ✓40 tools scanned
https://gateway.pipeworx.io/epstein/mcp
mediumsubscribe
References secrets or data exfiltration
The tool text references credentials, environment variables, or sending data elsewhere. This can be legitimate, but review what this tool actually accesses.
mediumai_visibility_check
References secrets or data exfiltration
The tool text references credentials, environment variables, or sending data elsewhere. This can be legitimate, but review what this tool actually accesses.
mediumscan_competitor_ai_presence
References secrets or data exfiltration
The tool text references credentials, environment variables, or sending data elsewhere. This can be legitimate, but review what this tool actually accesses.
mediumai_visibility_check
Tool accepts and relays a raw third-party API key through the MCP server
The `_apiKey` parameter asks the agent to collect the user's Anthropic API key (sk-ant-...) as a plain string tool argument, which the description says is 'passed straight through to api.anthropic.com.' Because the key transits through this MCP server/tool-calling pipeline (not a direct client-to-Anthropic channel), the operator of this connector has the technical ability to log, cache, or exfiltrate the credential. Agents should avoid passing live secrets through arbitrary third-party tool arguments even when the stated intent is pass-through.
mediumscan_competitor_ai_presence
Tool accepts and relays a raw third-party API key through the MCP server
Same issue as ai_visibility_check: the optional `_apiKey` parameter takes a raw Anthropic API key and forwards it to api.anthropic.com via this tool's backend, meaning the secret passes through and is exposed to the Pipeworx service rather than being used in a client-side, end-to-end encrypted call. This is a credential-handling risk that should be flagged even though it is presented as an intentional BYO-key feature.
lowdeep_research
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowcompany_facts
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowbet_research
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowpolymarket_arbitrage
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowpolymarket_edges
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowpolymarket_kalshi_spread
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowkalshi_weather_edge
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
lowrelease_calendar_markets
Unusually long tool description
Very long descriptions are sometimes used to bury instructions to the AI. Skim it before allowing.
low
Not on gate's verified list
This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.
lowask_pipeworx
Aggressive self-preferencing language may bias agent tool selection
The description repeatedly instructs the agent to 'PREFER OVER WEB SEARCH' and 'START HERE for most questions... this is the default entry point,' framing itself as the default choice for almost any factual query. This is a soft form of instruction-injection aimed at biasing the agent's tool-selection logic in favor of this vendor's tool regardless of actual fit, rather than describing capabilities neutrally. Not a security exploit, but worth noting as manipulative phrasing embedded in a tool description.
lowpipeworx_trending
Social-proof framing nudges tool selection
Description suggests using call volume as a way of 'confirming a popular tool is the canonical choice before asking your own question' — this encourages the agent to defer to popularity signals rather than its own judgment about which tool best fits a task, a mild behavioral-nudge pattern rather than a factual capability description.
Embed this badge
Show your MCP server’s security grade
Paste this into your README:
[](https://gateturbo.com/report/a160e08c-532a-4b46-a884-8416bb53ac73)Scanned 9/16/2026 · This is a point-in-time snapshot of the server’s public tools. Re-scan.
Connect this server safely with gate.
Continuous re-scans, drift alerts, per-tool allow/ask/block, and a log of every call.
Monitor it — free