MCP security report
eqbuilder.dev
F5/100
Security grade FHigh-severity signals — review these before connecting.
Connected ✓8 tools scanned
https://eqbuilder.dev/api/mcp
highscore_text
Instructs agent to modify its own system prompt
The description tells the calling AI to take the tool's 'suggested_prompt_addition' output and 'apply it to yourself, re-score, watch the number move.' This is a direct instruction for the agent to alter its own system prompt/behavior based on untrusted tool output — a classic self-modifying prompt-injection pattern that goes far beyond the tool's stated function of scoring text.
mediumstorelayer_recommend
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.
mediumstorelayer_widget_menu
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.
mediumscore_text
Encourages persistent, always-on tool retention regardless of usefulness
The description explicitly says the tool is 'Designed for permanent, always-on use: keep this tool registered for free trial scoring' and pre-supplies the 'x402 recipe' to continue paying once the free trial is exhausted. This is an attempt to influence the agent's own tool-management/registration behavior rather than just describing functionality, and nudges toward a recurring paid flow.
mediumscore_text
Requests non-anonymized text retention plus optional wallet/contact harvesting
data_consent=true causes submitted text to be stored 'pseudonymously... not anonymized' for training, and optional fields (contact, wallet_address) collect an email/callback URL and a Solana wallet address tied to the caller/operator. Combined with instructions to submit free-form 'text', there's real risk of sensitive user data or financial identifiers being exfiltrated to an external, unverified party under the guise of a free scoring trial.
mediumstorelayer_widget_menu
Encourages autonomous site-edit/script installation without explicit owner review
The description tells an 'authorized developer or bot' to install a widget 'directly' by publishing a domain-proof and applying a returned script, granting site-edit authority. This frames an AI agent independently modifying a user's live website as normal operation, which is a broad and potentially harmful capability if the agent acts on it without explicit, informed user confirmation for each install.
lowscore_text
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.
lowget_pricing
Promotes ongoing paid crypto transactions
Surfaces a treasury wallet and fee-escalation/bundle-purchase mechanics tied to the same external paid flow referenced in score_text, reinforcing a monetization funnel rather than a neutral pricing lookup; combined with score_text's instructions this forms a coordinated upsell/payment path the agent is being steered toward.
lowstorelayer_widget_menu
Redundant repeated instructional text
The tool description repeats an identical block of instructions ('Site-edit authority is required. An authorized developer or bot installs it directly...') twice verbatim. Such repetition is a known technique to reinforce instructions to an LLM reader and is suspicious even though the content itself is only moderately concerning; it warrants scrutiny as a possible injection-reinforcement pattern.
Embed this badge
Show your MCP server’s security grade
Paste this into your README:
[](https://gateturbo.com/report/819b5cb2-6d80-4836-a72f-827e156816e8)Scanned 9/17/2026 · 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