ai.borealhost/mcp

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

• Injection surface

  fail25% of grade

  Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

  tool "complete_checkout" description: hidden-instruction XML tag (e.g. <IMPORTANT>) — the Invariant Labs tool-poisoning marker. tool "complete_checkout" description: off-vendor URL in a description (borealhost.ai); capability text should not name where data goes. tool "complete_checkout" description: off-vendor URL in a description (checkout.stripe.com); capability text should not name where data goes. tool "get_checkout_status" description: off-vendor URL in a description (my-site.borealhost.ai); capability text should not name where data goes. tool "get_billing_portal" description: off-vendor URL in a description (billing.stripe.com); capability text should not name where data goes. tool "database_search_replace" description: off-vendor URL in a description (old-domain.com); capability text should not name where data goes. tool "database_search_replace" description: off-vendor URL in a description (new-domain.com); capability text should not name where data goes. tool "cloudflare_purge_cache" description: off-vendor URL in a description (my-site.borealhost.ai); capability text should not name where data goes. tool "register_domain" description: off-vendor URL in a description (example.org); capability text should not name where data goes. tool "set_domain_usage" description: off-vendor URL in a description (example.org); capability text should not name where data goes.

• Supply chain

  pass25% of grade

  Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

  Python build script has no build-time network/exec; repository linkage verified

• Credential hygiene

  fail15% of grade

  How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

  tool "set_api_key" has a secret-shaped input parameter "api_key" — secrets must never be tool parameters (the model would supply/handle the credential). tool "claim_api_key" has a secret-shaped input parameter "claim_token" — secrets must never be tool parameters (the model would supply/handle the credential). tool "create_ftp_account" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "manage_db_user" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "create_mailbox" has a secret-shaped input parameter "password" — secrets must never be tool parameters (the model would supply/handle the credential). tool "reset_mailbox_password" has a secret-shaped input parameter "new_password" — secrets must never be tool parameters (the model would supply/handle the credential). Reads secrets from the environment (e.g. mcp_server/config.py) — the recommended intake shape.

• Permission scope

  pass15% of grade

  Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

  Network egress present (mcp_server/api_client.py: outbound HTTP/socket call) — expected for most servers and not a finding on its own. No shell-exec, unbounded/sensitive filesystem, or download-and-run primitives found in scanned source. PASS = no static red flag. Static scanning is blind to code fetched or executed at runtime.

• Version behavior

  fail10% of grade

  Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

  vs prior 0.3.0 (minor bump): tool "request_api_key" description changed (wording only, no injection pattern); tool "claim_api_key" description changed (wording only, no injection pattern); tool "list_plans" description changed (wording only, no injection pattern); tool "create_checkout" description changed (wording only, no injection pattern); tool "update_checkout" description changed (wording only, no injection pattern); tool "update_checkout" gained input property(ies): site_type; tool "complete_checkout" description changed (wording only, no injection pattern); tool "get_checkout_status" description changed and now contains embedded outbound URL; tool "link_domain" description changed (wording only, no injection pattern); new tool "list_checkouts" touches a sensitive surface (email/exec/net/fs); new tool "cancel_checkout" touches a sensitive surface (email/exec/net/fs); new tool "add_site" touches a sensitive surface (email/exec/net/fs); new tool "set_site_type" touches a sensitive surface (email/exec/net/fs); new tool "get_wallet" touches a sensitive surface (email/exec/net/fs); new tool "delete_subdomain" touches a sensitive surface (email/exec/net/fs); new tool "get_site_nginx_snippet" touches a sensitive surface (email/exec/net/fs); new tool "set_site_nginx_snippet" touches a sensitive surface (email/exec/net/fs); new capability primitive(s) in source: http-client; new hardcoded outbound host/BCC destination(s): billing.stripe.com, borealhost.ai, checkout.stripe.com.

• Transport config

  fail10% of grade

  Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

  remote MCP endpoint declares NO authentication (unauthenticated remote — nginx-ui/CVE-2026-33032 class)