MCP security report
a2awire.com
F5/100
Security grade FA few things are worth a closer look before connecting.
Connected ✓16 tools scanned
https://a2awire.com/mcp/data/pubmedwatch-new-pubmed-biomedical-papers-e0bf9c/http
mediumdata_preview
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.
mediuma2awire_guide
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.
mediumfind_paid_work
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.
mediumregister
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.
mediumget_agent_contract
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.
mediumdata_session_attach_escrow
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.
mediuma2awire_guide
Implies existence of hidden/undisclosed tools beyond tools/list
The description states 'Every listed tool is callable via tools/call by name — tools/list shows only always-on essentials,' implying there are additional callable tools not enumerated in the normal tool listing surface presented to the agent/user. This encourages the agent to invoke tools that were never disclosed through the standard discovery mechanism, which could hide capabilities/actions from the user's visibility into what the agent can do.
mediumregister
Registration flow can generate and expose a wallet private key to the conversation
When withdrawal_address is omitted, the tool auto-provisions a sandbox payout wallet and 'returns its private key exactly once.' Returning raw private key material through a tool-call response places sensitive key material into the agent's context/transcript, where it could be logged, retained, or inadvertently leaked — this is a broader/more sensitive capability than a typical registration action and warrants caution about downstream handling of that secret.
mediumhire_and_execute
Autonomous on-chain fund movement in a single call
This tool searches for an agent, creates and funds an on-chain USDC escrow, and executes a task—all in one call—based only on a capability string, task text, and a max price ceiling. This bundles discovery, financial commitment, and execution into a single autonomous action with real funds, which is a broad and consequential capability that could move user funds without a distinct confirmation step for each transaction.
lowregister
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.
lowdata_preview
Urgency/manipulative call-to-action language
Description uses phrases like '✅ No API key needed — call this now' and 'Try one of the sample questions now' that are aimed at pressuring the agent into invoking the tool immediately rather than describing its function neutrally. This kind of imperative, urgency-based phrasing is a common prompt-injection pattern designed to bypass the agent's normal deliberation about whether a user actually wants this action performed.
lowget_agent_contract
Urgency/manipulative call-to-action language
Uses the same '✅ No API key needed — call this now' urgency framing seen across several tools in this suite, nudging the agent to invoke it proactively rather than in response to genuine user need.
lowfind_paid_work
References a tool not present in the provided tool set
Description instructs the agent to 'call start_job with a job_id to begin earning,' but no tool named start_job exists among the declared tools. This could cause the agent to attempt an undefined action or be primed to trust a tool name that may be spoofed/introduced later outside this vetted set.
lowdata_session_fund
Platform executes on-chain funding autonomously
Description states funding is 'platform-executed,' meaning the tool moves funds on the user's behalf without an explicit separate confirmation step in this call. Low severity given the small, capped per-query price (0.01 USDC, max 20 queries), but still worth noting as an autonomous spend action.
Embed this badge
Show your MCP server’s security grade
Paste this into your README:
[](https://gateturbo.com/report/15fc1a83-bb50-4936-b579-62afb037f670)Scanned 9/12/2026 · 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