All servers · Build & deploy

Novence

ai.novence/mcp · v0.1.3

Measured by MCP Station

Description written by the publisher, not verified

Hosted MCP for creating, checking, deploying, and hosting static sites for AI agents.

Connect

claude mcp add --transport http mcp https://api.novence.ai/mcp

{
  "mcpServers": {
    "mcp": {
      "url": "https://api.novence.ai/mcp"
    }
  }
}

Tools as reported by the server

• bootstrap
  Create an account + nv_ API key from an email. No Bearer key required. Unverified trial still publishes; OTP/verify_email is AFTER the live URL. Returns apiKey immediately — do not wait for OTP. If the human message includes a Signup attribution object, pass it as attribution unchanged.
• verify_email
  Confirm the email OTP (15 min TTL). No Bearer key required. Call after the first live URL, before checkout/mpp_upgrade.
• resend_verification
  Email a new OTP (also used for reissue_key). No Bearer key required. Works when already verified.
• reissue_key write/delete
  Mint a new account nv_ key from email OTP. No Bearer key required. WARNING: revokes prior account-scoped nv_ keys. Also verifies email if needed.
• checkout
  Return a Stripe Checkout URL for Pro ($29) or Scale. Requires nv_ + verified email. Use after a 2nd project or a 402 — never after the first live URL.
• mpp_upgrade
  Start or complete Pro/Scale via Machine Payments Protocol. Requires nv_ + verified email. First call returns a 402 Payment challenge; retry with payment_authorization (Payment credential) or _meta org.paymentauth/credential. Use after a 2nd project or a 402 — never after the first live URL.
• billing_portal
  Stripe Customer Portal URL for an existing subscriber (manage/cancel). Not first-time subscribe — use checkout or mpp_upgrade for that.
• create_project
  Create a new hosting project on the caller's account (counts against the caller's project quota). To edit a shared site, pass that project's project_id to upload/deploy tools instead.
• publish_html
  Publish one HTML document (report, dashboard, Claude-style artifact) as index.html and deploy. Omit project_id to create a project. Hosted MCP cannot read local files — pass the HTML string (read the file first). Returns url, project_id, and deployment. Poll get_deployment_status until live or faile
• list_projects
  List hosting projects the caller owns or is invited to (ids, names, suffixes). Use get_project for one project's settings; use get_preview_url for the public URL (per-deploy alias on Pro/Scale, live host on Free). Read-only; does not create a project.
• get_project
  Return one hosting project (name, suffix, settings, membership). Use list_projects to discover IDs. Use get_preview_url for the public URL (alias on Pro/Scale), and get_project_usage for quotas on that site. Read-only; does not mutate.
• update_project_settings
  Update project name, description, check thresholds, analyticsEnabled (cookieless edge traffic; off by default), shareGateEmails (optional OTP gate on platform hosts only; empty = public), siteMcp (off | read | forms visitor MCP on Pro/Scale), or owner-edit (data_edit_emails + data_edit_names, option
• get_upload_url
  Get a presigned PUT URL for one site file. content_type is optional (inferred from extension). Use get_upload_urls_batch for multi-file sites, then confirm_upload after the PUT.
• get_upload_urls_batch
  Presign many site files at once (max 100). Prefer this over get_upload_url for multi-file sites. After PUTs, call confirm_uploads_batch, not confirm_upload.
• confirm_upload
  Record one staged file after the client PUT to the presigned URL from get_upload_url. Use confirm_uploads_batch when several files were uploaded. Does not publish the site — call deploy after confirms. Re-confirming the same path updates metadata.
• confirm_uploads_batch
  Confirm many uploads after get_upload_urls_batch PUTs. 200 body is { files, errors }. errors is always an array (empty on full success); retry only failed paths. Use confirm_upload for a single file. Does not deploy.
• list_files
  List staged site files for a project (paths, types, sizes). Use get_file for one path's metadata; use delete_file to remove a path. Read-only; does not deploy.
• get_file
  Return metadata for one staged path (not file bytes). Use list_files to discover paths; use get_upload_url to replace content. Read-only.
• delete_file write/delete
  Remove one path from staging. Does not change the already-live deploy until the next deploy. Irreversible for that staged path; re-upload to restore. Use delete_form_submission for form PII, not this.
• set_redirects
  Write /_redirects (Netlify subset) into staging. Call deploy afterwards to publish. SPA History routing: /* /index.html 200. Exact files always win. Optional custom 404 is a separate /404.html upload. Does not write the live prefix.
• put_site_data
  Write /data/{name}.json on the live site without a deploy (Pro/Scale). Page JS keeps fetch("/data/{name}.json"). JSON must be an object or array, max 256 KB. Include the file in the next deploy or a repo upload that omits it can wipe it. Do not store secrets — the URL is public. Free: upload the JSO
• get_site_brief
  Private webmaster brief for this project (Pro/Scale). Markdown or JSON, max 64 KB. Not on the live origin. On session start call this then list_site_events. Empty is brief: null. Read-only.
• put_site_brief
  Write the private webmaster brief (Pro/Scale, max 64 KB). What the site is, which /data files matter, form ids, owner email, what not to touch. Empty string clears it. Survives deploys.
• list_site_events
  Webmaster inbox (Pro/Scale): form.submitted, collection.updated (agent vs owner_edit), check.completed, deploy.live. Default last 7 days, max 100. Pass since from the last session. Forms still email the owner. Read-only.
• deploy
  Publish the site. Verified accounts run quality checks; unverified accounts skip checks (checkRuns: 0) and still go live. POST returns quickly — poll get_deployment_status until live or failed. Do not retry POST while status is promoting or checking.
• get_deployment_status
  Poll publish state after deploy (queued, checking, promoting, live, failed). Omit deployment_id for the latest. Use get_checks_results for Lighthouse/a11y scores and get_preview_url once live. Read-only; do not retry deploy while promoting or checking.
• get_preview_url
  Return the public site URL. On Pro/Scale this is the per-deploy https://{suffix}--{ref}.novence.ai alias (last 5 successful deploys stay). On Free it is the live https://{suffix}.novence.ai host. Use get_deployment_status to know if live yet; use configure_custom_domain for a custom hostname. Read-o
• rollback
  Re-promote a previous successful snapshot to https://{suffix}.novence.ai without a new upload. Pro/Scale only (HTTP 402 on Free). Omit deployment_id to restore the previous live deploy. Only the last 5 successful deploys are retained. HTML Cache-Control is 60s — hard-refresh or wait. Does not run Li
• run_checks
  Start a Lighthouse, a11y, and link check cycle on the latest deployment. Does not publish a new deploy — use deploy for that. Then poll get_checks_results (not get_deployment_status) until scores appear. Unverified accounts skip checks on deploy; this still queues a cycle when allowed.
• get_checks_results
  Read Lighthouse, a11y, and link results for a deployment. Omit deployment_id for the latest. Use run_checks to trigger a new cycle; use get_deployment_status for publish/promote state, not scores.
• configure_custom_domain
  Attach a custom domain. For site domains, enables apex+www. Default primary=www. Required DNS: CNAME www → fallback.novence.ai. Then do exactly one apex option: ALIAS/ANAME/CNAME-flattening @ → fallback.novence.ai, or registrar URL-redirect @ → https://www.{domain}/. Pass primary=apex only when the
• get_domain_status
  Read DNS for the project's custom hostname after configure_custom_domain. Returns dns.www, dns.apex, and dns.next (a single next DNS step, or null when www CNAME is in place and apex has records). customDomainStatus stays pending after attach — poll dns.next until null. Do not use this to attach a d
• create_form
  Create a Novence form on a project (optional — sites may use Formspree/Web3forms instead). Returns submit URLs for edge + public API.
• list_forms
  List Novence forms on a project (ids, slugs, status). Use update_form to change one; use list_form_submissions to read leads. Not for third-party widgets. Read-only.
• update_form
  Patch an existing Novence form (name, slug, fields, notify_email, honeypot, active|disabled). Partial update: omitted fields stay unchanged. Use create_form for a new form; use list_forms to get form_id. Does not delete submissions.
• list_form_submissions
  List visitor submissions for one Novence form (paginated). Use delete_form_submission to remove PII. Not for update_form schema changes. Read-only.
• delete_form_submission write/delete
  Permanently delete one form submission (PII). Does not disable the form — use update_form status=disabled. Irreversible.
• get_quotas_and_usage
  Account-level plan limits and current consumption (projects, storage, deploys, bandwidth). Use get_project_usage for one site; use get_account for profile plus quotas together. Read-only snapshot, not a live meter stream.
• get_project_usage
  Per-project usage (storage, deploys, check minutes, bandwidth, forms). Bandwidth is live from edge-served bytes. Use get_quotas_and_usage for account totals; use get_project_analytics for pageviews. Read-only.
• get_project_analytics
  Cookieless edge analytics for a hosted site: pageviews, 404s, top pages, referrers, and countries. Off by default. Enable with update_project_settings analytics_enabled=true. Default last 7 days (max 90). If disabled, returns enabled=false and empty totals.
• get_account
  Full account snapshot: profile, subscription, quotas, usage, projects. Use get_quotas_and_usage for limits only; use get_account_console_kit to render a local HTML console. Read-only.
• create_account_session
  Mint a short-lived mgmt_ token for a local HTML console (account:read, billing:portal, project:members). Never put nv_ keys in HTML.
• get_account_console_kit
  JSON kit to write a local novence-console.html: snapshot, HTML template, session mint instructions. Serve on localhost. Use create_account_session for the mgmt_ token; use get_account for JSON without the kit. Read-only.
• invite_project_member
  Invite a collaborator by email. Requires Pro/Scale on the project owner. Only the owner can invite as admin. Creates a new project on the caller's account if you use create_project instead — pass project_id here. Do not share the owner's nv_ key.
• list_project_members
  List members and pending invites for a project. Use invite_project_member to add; use revoke_project_invite for pending tokens; use remove_project_member for accepted collaborators. Read-only.
• update_project_member
  Change a collaborator's role. Only the owner can grant or demote admin. Use remove_project_member to revoke access entirely. Does not affect pending invites — use revoke_project_invite for those.
• remove_project_member write/delete
  Remove an accepted collaborator and revoke their project API keys. Owner-only for admins. For a pending invite, use revoke_project_invite instead. Irreversible; they must be invited again.
• revoke_project_invite write/delete
  Cancel a pending invite so the token no longer works. Does not remove someone who already accepted — use remove_project_member for that. Irreversible for that invite_id; send a new invite_project_member if needed.
• leave_project write/delete
  Leave a project you were invited to. Owners cannot leave — transfer or stay. Revokes your project keys. Use remove_project_member if you are the owner removing someone else.
• accept_project_invite
  Accept an invite token (from email or invite_project_member in local/dev). Unverified invitees must pass otp. Returns a project-scoped nv_ key once — never email it. Collaborating uses this project_id; create_project still bills the caller.

Repository · Website · JSON · Text for agents · Registry updated 2026-09-05