MCP Tools Reference

This page documents the tool families exposed by the Link Platform MCP server. Use OAuth-based connector setup against https://{handle}.mcp.{domain}/mcp, with the shared-host compatibility path on https://ai.10x.in/mcp/{handle}/mcp only as a fallback when needed.

Hosted MCP can expose three capability shapes for a handle:

• tools for actions and structured reads
• read-only resources under tenx://...
• reusable prompts that package recommended workflows

This page focuses on tool calls. Use MCP Builder and Hosted Tools when you need the operator-facing browser workflow that reviews tools, resources, and prompts together.

Need the API + Edge + MCP flow in one place? Use Platform-Wide Feature Exploration Lab.

How tool responses work

Every tool returns the same envelope:

{
  "content": [{ "type": "text", "text": "<pretty-printed JSON>" }],
  "structuredContent": {
    /* same data as a plain object */
  }
}

On error:

{
  "isError": true,
  "content": [
    {
      "type": "text",
      "text": "{ \"error\": \"monolith_api_error\", \"status\": 403 ... }"
    }
  ],
  "structuredContent": {
    "error": "monolith_api_error",
    "status": 403,
    "body": { "error": "insufficient_scope" }
  }
}

For generic or local MCP clients, the handle parameter is optional on tools and falls back to the server's LINK_PLATFORM_DEFAULT_HANDLE (set at session creation time via the subdomain, e.g. hello.mcp.arjun.tv).

For ChatGPT/OpenAI hosted connectors, do not pass handle when the connector is already bound to one handle. One connector now maps to one handle, and the hosted MCP runtime will use that bound handle automatically.

---

Scope quick reference

The live tool catalog appears in tools/list regardless of scopes. Scope errors only surface when a tool is actually called.

Some handles also expose pack-specific resources and prompts. For example, commerce-enabled handles may expose resources such as tenx://commerce/overview and guided prompts such as refund_triage alongside the tool catalog.

Companion workflow guide:

• Platform-Wide Feature Exploration Lab

---

Links

links_list

List all short links for a handle. Returns each link's slug, destination URL, title, enabled status, and health status (OK, BROKEN, or UNREACHABLE).

Inputs

Example prompt

"List all my links for handle hello."

Response shape

{
  "handle": "hello",
  "links": [
    {
      "slug": "pricing",
      "destinationUrl": "https://example.com/pricing",
      "title": "Pricing page",
      "enabled": true,
      "position": 1,
      "createdAt": "2026-01-10T12:00:00Z"
    }
  ]
}

---

links_upsert

Create a new short link or update an existing one. Provide a slug (the URL path), destinationUrl, and optional title. Use this for quick link creation without form validation — use links_form_submit if you need schema-based validation.

Inputs

Example prompt

"Create a link with slug launch pointing to https://example.com/launch for handle hello."

Response shape

{
  "slug": "launch",
  "destinationUrl": "https://example.com/launch",
  "title": "Launch",
  "enabled": true,
  "updatedAt": "2026-03-02T09:00:00Z"
}

---

links_health_check

Check if link destination URLs are reachable. Visits each link's destination and reports OK, BROKEN (e.g. 404), or UNREACHABLE.

Inputs

Example prompt

"Run a health check on all links for hello and tell me which ones are broken."

Response shape

{
  "results": [
    { "slug": "pricing", "status": 200, "healthy": true },
    { "slug": "old-page", "status": 404, "healthy": false }
  ]
}

---

links_route_preview

Preview where a short link would redirect to under specific visitor conditions (country, device, source, campaign). Useful for testing personalization rules without actually clicking the link.

Inputs

Example prompt

"Show me where the pricing link would redirect a mobile visitor from the US."

Response shape

{
  "slug": "pricing",
  "resolvedUrl": "https://example.com/pricing-mobile",
  "ruleApplied": "mobile-us-variant",
  "fallback": false
}

---

Site Management

site_deploy_inline

Deploy one small HTML-only page directly. Best for landing pages and simple sites without separate JS/CSS/image files.

When to use this vs site_deploy_multifile:

• Use site_deploy_inline for one HTML-only page with optional inline <style> tags. Script tags (<script>) and inline event handlers (onclick, etc.) are blocked for security.
• Use site_deploy_multifile for real websites, routed pages, separate JavaScript/CSS/image files, or edits to an already deployed site.

Inputs

Example prompts

"Deploy this HTML as my landing page." "Publish this one-page HTML promo at /summer-sale."

Response shape

{
  "handle": "hello",
  "deploymentId": "dep_abc123",
  "status": "ACTIVE",
  "liveUrl": "https://hello.10x.in/",
  "pageUrls": ["https://hello.10x.in/"],
  "activatedAt": "2026-03-02T09:05:00Z",
  "previewUrl": "https://..."
}

---

site_deploy_multifile

Deploy a full website with separate HTML, CSS, JS, and image files in a single tool call.

This is the default tool for real websites, routed pages, and edits to an already deployed site.

Use this when the site needs JavaScript files, stylesheets, or images as separate files — unlike site_deploy_inline which only accepts inline HTML without script tags. Use nested index.html files to publish specific public paths. Examples: courses/index.html, store/index.html, and dashboard/index.html publish /courses, /store, and /dashboard. perfume/index.html becomes https://hello.10x.in/perfume.

How it works:

The tool supports two modes depending on whether you provide file content:

Mode 1: Automatic (recommended for AI clients) — Provide content for each file. The MCP server handles everything: creates a draft, uploads all files to storage, and activates the deployment. Returns a live URL. This is the recommended mode for ChatGPT, Claude, and all AI clients.

Mode 2: Manual upload — Omit content from files. The tool returns pre-signed S3 upload URLs. You upload files yourself via HTTP PUT, then activate separately. This mode is for developer scripts, CI/CD pipelines, and binary file deployments.

Inputs

File object fields:

Example prompts

"Deploy my website with index.html, styles.css, and app.js" "Create a site with a home page, about page, and contact page with JavaScript interactivity"

Response shape (Mode 1 — automatic, with content)

{
  "handle": "hello",
  "deploymentId": "dep_xyz789",
  "status": "ACTIVE",
  "liveUrl": "https://hello.10x.in/",
  "pageUrls": ["https://hello.10x.in/", "https://hello.10x.in/about"],
  "uploadResults": [
    { "path": "index.html", "status": "uploaded" },
    { "path": "styles.css", "status": "uploaded" },
    { "path": "about/index.html", "status": "uploaded" }
  ]
}

Response shape (Mode 2 — manual, without content)

{
  "handle": "hello",
  "deploymentId": "dep_xyz789",
  "status": "DRAFT",
  "liveUrl": "https://hello.10x.in/",
  "pageUrls": ["https://hello.10x.in/", "https://hello.10x.in/about"],
  "uploadUrls": [
    {
      "path": "index.html",
      "contentType": "text/html",
      "putUrl": "https://s3.amazonaws.com/..."
    },
    {
      "path": "styles.css",
      "contentType": "text/css",
      "putUrl": "https://s3.amazonaws.com/..."
    },
    {
      "path": "about/index.html",
      "contentType": "text/html",
      "putUrl": "https://s3.amazonaws.com/..."
    }
  ]
}

Manual activation (Mode 2 only): After uploading files to the pre-signed URLs, activate with:

POST /v2/public/handles/{handle}/site-deployments
Content-Type: application/json

{ "activate": true, "deploymentId": "dep_xyz789" }

Limits:

Mode 1 (with content) works for text-based files: HTML, CSS, JS, JSON, SVG, TXT. For binary files like PNG/JPG images, use Mode 2 or reference images by external URL in your HTML (e.g. ``).

---

site_release_get_files

Read files from the current live release by default, or from a specific release when releaseId is provided.

Use this as the read step before editing a deployed site with AI. The tool returns inline content for editable text files and metadata-only entries for binary assets.

Inputs

Behavior

• Text files such as index.html, styles.css, app.js, data.json, README.md, and icon.svg return inline content
• Binary files such as .png, .jpg, .woff, and .pdf return metadata only
• If the combined text content would be too large, some files return contentOmittedReason: "response_too_large" instead of full content

Example prompt

"Read the current live release files for my site so you can edit the hero section."

Response shape

{
  "handle": "hello",
  "releaseId": "dep_live123",
  "isLive": true,
  "fileCount": 3,
  "files": [
    {
      "path": "index.html",
      "contentType": "text/html",
      "sizeBytes": 1824,
      "isText": true,
      "content": "<!DOCTYPE html>..."
    },
    {
      "path": "assets/app.js",
      "contentType": "application/javascript",
      "sizeBytes": 624,
      "isText": true,
      "content": "console.log('hello')"
    },
    {
      "path": "assets/logo.png",
      "contentType": "image/png",
      "sizeBytes": 20480,
      "isText": false
    }
  ]
}

---

site_release_rollback

Restore an older site release so it becomes live again.

Use this when a new site deployment was bad and you want to bring back a previous working version quickly.

Inputs

Example prompt

"Rollback my site to release dep_prev_good."

Response shape

{
  "handle": "hello",
  "releaseId": "dep_prev_good",
  "status": "LIVE",
  "liveReleaseId": "dep_prev_good",
  "rolledBackFromReleaseId": "dep_bad_new"
}

site_list_deployments

List all site deployments for a handle.

In ChatGPT, successful results render a launcher button for the live site URL.

Inputs

Example prompt

"List recent deployments for hello."

Response shape

{
  "handle": "hello",
  "activeDeploymentId": "dep_abc123",
  "liveUrl": "https://hello.10x.in/",
  "deploymentCount": 2,
  "deployments": [
    {
      "deploymentId": "dep_abc123",
      "status": "ACTIVE",
      "isActive": true,
      "liveUrl": "https://hello.10x.in/",
      "createdAt": "2026-03-02T09:05:00Z",
      "activatedAt": "2026-03-02T09:05:00Z"
    },
    {
      "deploymentId": "dep_old456",
      "status": "INACTIVE",
      "createdAt": "2026-02-20T14:00:00Z"
    }
  ]
}

---

site_preview_deployment

Get a signed preview URL for any deployment (active or draft).

In ChatGPT, successful results render launcher buttons for the signed preview URL and the live site URL.

Inputs

Example prompt

"Give me a preview link for deployment dep_abc123 on handle hello."

Response shape

{
  "handle": "hello",
  "deploymentId": "dep_abc123",
  "liveUrl": "https://hello.10x.in/",
  "previewUrl": "https://ai.10x.in/v2/public/preview/activate?token=eyJ...",
  "expiresInSec": 900
}

---

pages_list

List creator pages for a handle with an optional status filter.

In ChatGPT, successful results render launcher buttons for the returned public page URLs.

Inputs

Example prompt

"Show me all published pages for hello."

Response shape

{
  "handle": "hello",
  "pageCount": 1,
  "pages": [
    {
      "pageSlug": "welcome",
      "title": "Welcome",
      "status": "PUBLISHED",
      "pageUrl": "https://hello.10x.in/welcome"
    }
  ]
}

---

campaign_structure_get

Fetch campaign detail structure for a campaign (CAMPAIGN_DETAIL_V1 pages).

Inputs

Example prompt

"Fetch the campaign structure for campaign cmp_launch on handle hello."

Response shape

{
  "handle": "hello",
  "campaignId": "cmp_launch",
  "pageSlug": "campaign-launch",
  "accessMode": "FREE_WITH_LOGIN",
  "status": "PUBLISHED",
  "structureType": "CAMPAIGN_DETAIL_V1",
  "schemaVersion": 3,
  "schemaS3Key": "site-structure/hello/campaigns/cmp_launch/v3.json",
  "schema": {
    "modules": [{ "id": "hero", "type": "hero_banner" }]
  }
}

---

campaign_structure_upsert

Create or update the campaign detail structure page and schema JSON.

Inputs

Example prompt

"Upsert campaign structure for cmp_launch with pageSlug campaign-launch and schema modules."

Response shape

{
  "ok": true,
  "handle": "hello",
  "campaignId": "cmp_launch",
  "pageSlug": "campaign-launch",
  "accessMode": "FREE_WITH_LOGIN",
  "status": "DRAFT",
  "schemaS3Key": "site-structure/hello/campaigns/cmp_launch/v4.json",
  "schemaVersion": 4
}

---

campaign_structure_publish

Publish campaign detail structure page (server enforces FREE_WITH_LOGIN).

Inputs

Example prompt

"Publish campaign structure page for cmp_launch."

Response shape

{
  "ok": true,
  "handle": "hello",
  "campaignId": "cmp_launch",
  "pageSlug": "campaign-launch",
  "accessMode": "FREE_WITH_LOGIN",
  "publishedAt": "2026-03-06T12:00:00.000Z"
}

---

Knowledge And Context Store

knowledge_query

Search manually managed and indexed handle knowledge using natural language. Use this for approved documents, runbooks, policies, and manually authored context.

Inputs

context_store_search

Search or list replicated connector records. Use this for read-only connector data when the source/entity status is READY or PREVIEW.

Inputs

Example prompt

"Check Context Store for project proj_launch and find repository docs about deployment validation."

Response shape

{
  "executionPath": "context_store",
  "query": "deployment validation",
  "count": 2,
  "results": [],
  "sourceStatuses": [
    {
      "sourceId": "repo_docs",
      "connectorKey": "github_docs",
      "status": "READY",
      "lastSyncAt": "2026-05-06T10:00:00Z"
    }
  ]
}

context_store_status

Report Context Store source and entity readiness, counts, cursors, and timestamps.

Inputs

context_source_sync

Request an async refresh for a configured Context Store source. Use direct provider calls instead when writes or real-time freshness are required.

Inputs

---

Companion API/Edge Surfaces (Function Bindings)

Function binding creation and runtime invocation are API/Edge flows that complement MCP usage.

Create function binding (control-plane)

POST /v2/handles/{handle}/function-bindings

Bindings now declare:

• status: ACTIVE or DISABLED
• target: template or skill
• exposure.publicWeb and exposure.authenticatedMcp
• execution.mode: sync or async
• tool.name, tool.description, and inputSchema
• config: target-specific runtime config

Public Web exposure is limited to template-backed bindings in v1.

Discover enabled bindings (public handle)

GET /v2/public/handles/{handle}/function-bindings

Returns only ACTIVE template-backed bindings exposed on the public Web lane.

Discover authenticated MCP bindings

GET /v2/public/handles/{handle}/function-bindings/mcp

Requires PAT scope mcp.connect plus binding/skill invoke authorization.

Invoke binding runtime (edge)

GET|POST /_edge/fn/{handle}/{bindingKey}

Use this for request-level invocation tests and integration checks.

Invoke authenticated MCP binding

POST /v2/public/handles/{handle}/function-bindings/{bindingKey}/invoke

Returns the normalized sync or async tool envelope used by hosted/compat MCP.

For runnable examples, use Platform-Wide Feature Exploration Lab.

---

Analytics

analytics_get

Get click and conversion analytics. Shows total clicks, conversion rates, and traffic data. Use groupBy='slug' to see per-link stats, or 'country'/'device'/'referrer' for breakdowns. Supports date ranges with fromDay/toDay (YYYY-MM-DD).

Inputs

Example prompt

"How many clicks did handle hello get in the last 7 days?"

Response shape

{
  "handle": "hello",
  "period": "7d",
  "clicks": 1842,
  "conversions": 94,
  "conversionRate": 0.051
}

---

analytics_export

Export raw analytics data for reporting or external tools. Returns the full dataset for the specified date range.

Inputs

---

analytics_campaign_health

Check health signals for a specific campaign — click trends, conversion rates, and anomalies.

Inputs

---

Tracking

tracking_list_templates

List all tracking integrations (Meta Pixel, Google Tag Manager, GA4, custom webhooks) and their enabled/disabled status.

Inputs

Response shape

{
  "templates": [
    {
      "templateKey": "meta_pixel",
      "enabled": true,
      "config": { "pixelId": "12345" }
    },
    {
      "templateKey": "ga4_event",
      "enabled": true,
      "config": { "measurementId": "G-XXXX" }
    }
  ]
}

---

tracking_upsert_template

Add or update a tracking integration. Supported types: meta_pixel (Facebook), gtm_event (Google Tag Manager), ga4_event (Google Analytics 4), custom_webhook.

Inputs

templateKey must be one of: meta_pixel, gtm_event, ga4_event, custom_webhook. This tool does not accept freeform {{variable}} UTM string templates.

Example prompt

"Add a Meta Pixel with ID 98765 to handle hello."

Working example

{
  "handle": "deepak",
  "payload": {
    "templateKey": "ga4_event",
    "enabled": true,
    "config": {
      "eventName": "ai_masterclass_visit"
    }
  }
}

---

tracking_list_personalization_rules

List all A/B testing and personalization rules. Shows each rule's conditions (country, device, source), variant, priority, and traffic allocation.

Inputs

---

tracking_upsert_personalization_rule

Create or update an A/B testing or personalization rule. Target visitors by country, device, traffic source, campaign, or segment. Each rule has a variant (with required id, optional title/cta/destinationUrl) and optional traffic allocation (0-1). Lower priority numbers run first.

Inputs

There are no generic operators such as equals. Matching uses the fixed ...In arrays inside payload.when.

Working example

{
  "handle": "deepak",
  "ruleId": "ai-masterclass-2026-landing",
  "payload": {
    "enabled": true,
    "priority": 100,
    "when": {
      "campaignIn": ["ai-masterclass-2026"]
    },
    "variant": {
      "id": "masterclass_variant_a",
      "title": "AI Masterclass 2026",
      "cta": "Reserve your seat",
      "destinationUrl": "https://deepak.10x.in/masterclass"
    }
  }
}

---

tracking_resolve_context

Resolve a ?ctx= token into its attribution data and variables. No connector token required — public endpoint.

Inputs

Response shape

{
  "attribution": { "source": "newsletter", "campaign": "march-launch" },
  "vars": { "variant": "B", "segment": "returning" }
}

---

tracking_write_signal

Write a behavioral signal into the chain for a visitor session.

Inputs

---

tracking_resolve_chain

Evaluate a chain trigger and get the routing decision for a visitor.

Inputs

---

Routing

routing_list_context_origins

List allowlisted browser origins that can generate CTX tokens.

Inputs

---

routing_update_context_origins

Replace the full list of allowlisted browser origins.

Inputs

---

routing_read_chain_session

Read a chain session — accumulated signals and current state.

Inputs

---

routing_prefetch_decisions

Pre-compute routing decisions for multiple trigger events in a single call.

Inputs

---

Webhooks

webhooks_list

List all webhook subscriptions for a handle.

Inputs

Response shape

{
  "subscriptions": [
    {
      "subscriptionId": "wh_abc",
      "endpointUrl": "https://myserver.com/hook",
      "eventTypes": ["link.clicked", "conversion.recorded"],
      "enabled": true
    }
  ]
}

---

webhooks_create

Register a new webhook subscription.

Inputs

Example prompt

"Create a webhook for handle hello that fires on link.clicked events and sends to https://myserver.com/hook."

Working example

{
  "handle": "deepak",
  "payload": {
    "endpointUrl": "https://n8n.1to10x.ai/webhook/lead-capture",
    "eventTypes": ["lead.captured", "link.clicked"],
    "enabled": true
  }
}

The destination URL does not need pre-registration in 10x. It must be public HTTPS, resolvable, and outside the blocked private-host list.

Supported event names include link.clicked, deployment.created, deployment.activated, skill.invoked, link.routed, link.context_created, conversion.recorded, personalization.resolved, campaign.created, campaign.health_changed, lead.captured, moderation events, webhook.delivery.terminal_failure, destination-domain events, groupbuy events, scene events, agent events, and QA events.

---

webhooks_delete

Delete a webhook subscription by ID.

Inputs

---

webhooks_test

Send a test event to a webhook to verify delivery.

Inputs

---

Agent (AI Planning & Execution)

The agent tools follow a lifecycle: discover → strategy → proposal → run.

agent_discover

Run agent discovery to surface marketing signals and opportunities for a handle.

Inputs

---

agent_generate_strategy

Generate strategy recommendations based on discovery signals.

Inputs

---

agent_list_proposals

List execution proposals, optionally filtered by status (PENDING, APPROVED, REJECTED, RUNNING, COMPLETED).

Inputs

---

agent_create_proposal

Create an execution proposal from strategy output.

Inputs

---

agent_approve_proposal

Approve a pending proposal so it can be executed.

Inputs

---

agent_reject_proposal

Reject a pending proposal with an optional reason.

Inputs

---

agent_start_run

Start execution of an approved proposal.

Inputs

---

agent_get_run_status

Poll the status of a running or completed execution, including step-level details.

Inputs

Response shape

{
  "runId": "run_xyz",
  "status": "RUNNING",
  "steps": [
    { "stepId": "s1", "name": "upsert-links", "status": "COMPLETED" },
    { "stepId": "s2", "name": "deploy-site", "status": "RUNNING" }
  ]
}

---

agent_rollback_run

Rollback a completed run, reverting changes made during execution.

Inputs

---

System

system_health

Check if the MCP server is connected and working. Returns ok: true if the platform is reachable, plus a count of links as a health probe.

Inputs

Example prompt

"Is the MCP connection healthy for handle hello?"

Response shape

{
  "handle": "hello",
  "healthy": true,
  "links": 12
}

---

system_usage_meters

Check current API usage and rate-limit status. Shows how much of your quota has been used.

Inputs

---

system_audit_events

View recent activity log — shows who created links, deployed sites, deleted webhooks, and other actions taken on this handle.

Inputs

---

Code Runs

code_runtime_list

List configured code runtimes available for a handle.

Inputs

---

code_run_submit

Submit a code run with an idempotency key.

Inputs

---

code_run_status

Fetch the current status of a submitted code run.

Inputs

---

code_run_fetch_output

Fetch inline output or file output metadata for a completed run.

Inputs

---

Forms

forms_schema_get

Get the form schema that defines what fields are available when creating links (field names, types, required/optional). Currently supports link_form_v1.

Inputs

link_form_v1 is the shipped link-management baseline schema. There is currently no built-in lead-capture baseline form for fields like name, email, phone, or job title.

---

forms_feedback_record

Record feedback about a form submission to improve the form schema over time. Send outcome: 'success' with a usedElements array listing which fields were used.

Inputs

---

links_form_submit

Create or update a link with form validation. First fetches the form schema, validates required fields (like destinationUrl), then creates/updates the link. Use this instead of links_upsert when you want schema-validated link creation.

Inputs

---

Troubleshooting

Related

• MCP Builder and Hosted Tools
• ChatGPT MCP Setup
• API Tokens and Automations
• Apps and MCP Tools
• API Auth and Error Model