Fermat Platform (External)

Descriptor

Candidate descriptor · transport streamable-http (POST); GET returns 405 with a JSON-RPC advisory that this server is POST-only

Tools

• rank_products — Rank products by how well they match user preferences or quiz answers. Uses deterministic attribute matching — no LLM call, runs in <1ms. Pass products (from search_products_es with include_attrs=true) and a preferences object mapping attribute keys to desired values. Returns products sorted by match score with a _score field added. Example preferences for a bike quiz: {"terrain": "mountain", "skill_level": "intermediate", "activity": "trail riding"} Example preferences for waterproof pants: {"material": "GORE-TEX", "activity": "skiing", "weight_class": "lightweight"}
• search_products_es — Search products using Elasticsearch with full-text + semantic search, category filtering, attribute facets, and price ranges. Much faster and more relevant than the Hasura-based search_products. Returns product name, handle, price, image, brand, and optionally attributes. Filter DSL examples: - "category:Bikes>Mountain" — category prefix - "price:<=500" — price ceiling - "price:>=100,price:<=500" — price range - "brand:Cannondale" — brand filter - "category:Bikes,price:>100,brand:Trek" — combined Set include_attrs=true to get attribute facets for product intelligence scoring.
• get_ais_pending_funnel_layouts — Get AIS (AI Search) funnel layouts for an organization, filtered by review status. Returns the full layout JSON (all modules and their content/copy) for each matching funnel. Defaults to pending funnels. Use this to load funnel content for compliance validation against brand guidelines. Each funnel includes shop_id, tile_id, review metadata, and the complete module layout.
• get_ais_brand_guidelines — Get all brand guideline content for an organization: Knowledge Base documents (pre-converted markdown from GCS, preserving tables and document structure), Pierre brand voice, and Pierre extra context. Returns the full text of every KB document grouped by source file, plus brand voice and extra context fields. Use this to load the complete set of brand rules, compliance guidelines, trademark requirements, and tone directives before validating AIS funnel copy.
• get_custom_module — Get a custom module's full details including the generated HTML/CSS/JS bundle code, config schema, and default config. Look up by module_name (e.g. 'custom/bike-finder-quiz') or by id (UUID). The generated_component field contains the full bundle source.
• list_custom_modules — List custom modules in an organization's module library. Returns module name, display name, description, status, and timestamps. Use get_custom_module to fetch the full bundle code. Supports optional name search and pagination. By default only lists `ready` modules. Pass `include_non_ready: true` to also include in-flight (`generating`) and failed (`failed`) generation jobs.
• list_forge_artifacts — List artifacts (documents, images, JSON files) produced by a Forge workflow run. Returns each artifact's name, type, content type, storage URL, and review status. Use this to find downloadable outputs after a run completes.
• get_forge_run_outputs — Fetch the outputs of a completed Forge workflow run, including the final run output and per-node execution details. Each node output includes the node_id, node_type, output data, execution order, duration, and status. Use this after a run completes to inspect what each workflow step produced.
• get_forge_run_status — Check the status of a Forge workflow run by its run_id. Returns the current status (pending, running, completed, failed), the current node being executed, any error message, and timing information. Use this to poll a run after starting it with start_forge_run.
• list_forge_templates — List published Forge workflow templates for an organization. Returns each template's name, slug, description, and input_schema so you know what inputs are required to start a run. Use the input_schema to understand the expected shape of the `input` parameter for start_forge_run.
• list_filter_settings — List product attribute filters for an organization. By default returns a summary of all attribute keys with value counts and their storefront sort position (display_order). Pass key_name to drill into a specific key and see all its values with their visibility and display name settings.
• get_bundle — Get a bundle's full details including products, images, layout, linked tiles, and BYOB config (if applicable). Use list_bundles first to find bundle IDs.
• get_virtual_product — Get a virtual product's full configuration including all member products with their variants, images, option value overrides, and clustering settings.
• list_virtual_products — List all combined/virtual products for an organization. Virtual products merge multiple real products into a single product page with shared variant selection. Returns each VP's config: key product, member products, option overrides, and clustering mode.
• list_pdp_versions — List all PDP versions for a root product. Returns each version's ID, name, handle, and whether it is the core or favorite (default) version. Also shows whether each version has a custom PDP layout. Use search_products first to find the root_product_id.
• get_offers_for_shop — Get all offers configured on a shop (via its default tile). Returns the shop name, tile info, and all offer configs including discount types, tiers with discount codes, and NoOp status. Accepts a shop ID (UUID from the dashboard URL or list_funnels output).
• list_offers — List all offers for an organization. Returns offer details including discount type (order, product, shipping), threshold type, tiers with discount codes/values, and which tiles use each offer. Useful for debugging discount code issues and comparing offer configs across funnels. Supports filtering by active status and search by name.
• list_recommendation_exclusions — List products excluded from recommendation algorithms for an organization. Returns product names, handles, categories, who excluded them, and when. Set summary_only=true to get just the total count and breakdown by category without individual product details — useful for orgs with many exclusions.
• get_product_funnel_layout — Get a product funnel layout by ID. Returns the full record including both lander_layout and pdp_layout. Use this as a starting point before creating a product funnel layout preview session.
• list_product_media — List media files associated with products for an organization. Shows images from Shopify product catalog, user uploads, and other sources. Optionally filter by product_id to get media for a specific product. Each result includes the media URL, source platform, and associated product ID.
• list_org_media — List media files from an organization's media library. Returns image and video URLs from the org's uploaded assets (excludes fonts and auto-pulled Facebook media). Use this to find images for swapping into module layouts. Each result includes the full CDN URL ready to use in module fields.
• get_shop_layout — Get the current (latest version) layout JSON for a tile/shop. Returns the full layout structure including all modules and their configuration. Use this as a starting point before creating a shop layout preview session. The response includes `updated_at`, an etag-style token. To safely persist edits with save_shop_layout you MUST pass this value back as `expected_updated_at` so concurrent edits made elsewhere (e.g. via 'Edit With AI' in the dashboard) are not silently overwritten.
• list_modules — List all available Fermat modules from the module registry, grouped by category. Each module includes its moduleName, displayName, category, and description. Use the optional 'category' filter to narrow results (e.g. 'Commerce', 'Content').
• list_product_funnel_templates — List available product funnel templates (layouts) for an organization. Returns each template's ID, name, and timestamps. Use this to discover valid product_funnel_layout_id values when creating a product group. Supports pagination via limit/offset.
• list_product_groups — List product groups for a product funnel. Returns each group's ID, name, description, precedence order, associated funnel and layout names, member count, and timestamps. Supports pagination via limit/offset.
• get_product_info — Get comprehensive information about a product. Returns the full product model including all variants (with pricing), images, PDP layouts (both authored and resolved), product versions, layout fragment keys, and product group membership. Look up by root_product_id, or by org_id + handle.
• get_batch_fragment_status — Lightweight status check for batch fragment generation. Returns aggregate counts (total_members, with_fragment, without_fragment, disabled_products) and a per-member breakdown (root_product_id, name, external_id, has_fragment, externally_enabled). Use before batch generation as a dry run and after generation to verify results.
• list_product_group_members — List all products in a product group with enriched data. Returns per-member: root_product_id, core_product_id, external_id (SKU), name, handle, description, image_url, externally_enabled. Optionally include fragment_key to check existing fragment status per product ({exists, updated_at}), and include_tct_keys to batch-fetch tile config tag values ({available, value} per key). Supports pagination via limit/offset. Batch-fetches in 1-3 Hasura queries instead of N individual calls.
• get_product_groups_for_product — Given a root_product_id, discover which product group(s) the product belongs to. Returns an array of groups with ID, name, funnel name/slug, layout name, precedence order, and member count. An empty array means the product is not in any product group. Use this as the first step before batch-generating layout fragments.
• get_brand_voice — Get brand voice and extra context for an organization. Returns the brand's personality traits and additional context that should guide tone, style, and content generation. Use this to load brand voice into your context before generating copy, product descriptions, or any brand-specific content.
• get_org_branding — Get branding settings for an organization — colors (primary, secondary, accent, button colors), fonts (primary, secondary, up to 3 custom fonts with variants), button styles (shape, padding, font config, custom CTA text), logo, and favicon. Use this before editing layouts to ensure modules use the correct brand colors, fonts, and button styling.
• get_layout_fragments — Get layout fragments for a root product. Layout fragments are reusable pieces of layout configuration stored per product. Filter by fragment_key to get a specific fragment.
• get_bundle_layout — Get the current BDP (Bundle Detail Page) layout for a bundle. Returns the bundle's layout JSON along with bundle metadata (name, type, org_id). Use this to read existing bundle layouts before making edits.
• list_bundles — List all bundles for an organization, including their products and pricing. Returns bundle name, type, discount, and included products with quantities.
• get_layout_version — Get the full layout JSON for a specific version of a tile. Returns the complete layout structure including all modules and their configuration. Requires tile_id and version_number. Get version numbers from get_layout_history.
• get_layout_history — Get the version history of a tile's layout, showing when changes were made and by whom. Use this to understand how a funnel's design has evolved. Requires tile_id. Returns version_number, created_at, created_by, and message for each version.
• search_products — Search products in the catalog. Returns lightweight results (id, external_id, name, handle, externally_enabled, available_variants count). Use get_product_info with a root_product_id to fetch full details (images, variants, categories, layouts) for specific products. Can filter by name, SKU (external_id on root_products), category (matched as a prefix against pipe-delimited normalized_path — e.g. passing 'Accessories' finds products under 'Accessories', 'Accessories|Books', etc.), and whether they're externally enabled (live on the store). Defaults to only returning in-stock products (has at least one available variant); set in_stock=false to include out-of-stock.
• search_organizations — Search for organizations (brands) by name. Returns the org ID needed by all other tools, plus metadata like website, ecommerce platform, vertical, and account type. Use 'search' for name filtering. Supports pagination.
• get_friction_point_report — Generate a friction point detection report for a brand's website. Fetches all behavioral anomalies (rage clicks, dead clicks, confusion loops, u-turns, excessive scrolling) and generates a detailed AI report for each one with session recording links, root cause analysis, and recommendations. Combines get_anomalies + get_anomaly_report into one call. Supports all anomaly types: behavioral (rage clicks, dead clicks, confusion loops), traffic, conversion, and search. Defaults to behavioral if not specified. All types include session recording links. Date range defaults to last 7 days if omitted.
• get_anomaly_report — Get a detailed AI-generated report for a specific anomaly, including root cause analysis and recommendations. Requires anomaly_id and anomaly_type from get_anomalies results.
• get_anomalies — Get recent anomalies detected by Fermat's anomaly detection system. Returns behavioral, traffic, conversion, and search anomalies with details. Filter by anomaly_type (behavioral/traffic/conversion/search), date range, and limit.
• ask_pierre — Ask Pierre, Fermat's AI assistant, a question. Pierre has access to BI analytics, shop creation, bundle creation, behavioral anomaly analysis, module customization, and more. Pierre automatically routes your query to the appropriate sub-agent.
• list_active_dpp_experiments — List all currently live DPP (product funnel) experiments across ALL organizations in a single call. Returns experiment metadata grouped by org — org name, experiment name, start date, and layout variants. Use this for cross-org discovery (e.g. daily significance monitoring) instead of iterating list_experiments per org.
• get_external_session_journey — Get detailed session-level data from the brand's OWN website (external site, not Fermat funnels). Each row is one session with: user path (page sequence), entry/exit pages, engagement signals (product views, add-to-cart count, checkout initiated), behavioral metrics (rage_click counts, etc.), and intention score. Returns multiple sessions by default (up to limit). Use session_id to drill into one specific session. Date range is optional — defaults to the last 30 days if omitted.
• get_external_session_behavioral_insights — Get aggregate behavioral patterns from session recordings on the brand's OWN website (external site, not Fermat funnels). Shows how external sessions break down by intention score, device type, traffic source, and entry page type, with engagement metrics (avg duration, scroll depth, clicks, product views, time per page). Compare with get_session_behavioral_insights (Fermat funnels) to understand how user behavior differs between the brand site and Fermat experiences. Filter by behavioral tag to focus on UX problems. Date range is optional — defaults to the last 30 days if omitted.
• get_session_checkout_analysis — Analyze checkout conversion patterns across session recordings. Shows where checkout conversions drop off by segment: traffic source, device type, intention score, and country. Returns session counts, checkout completion counts, revenue, and engagement metrics for each segment. Use this to answer questions like 'which traffic sources convert best?' and 'do mobile users check out less than desktop?'. Date range is optional — defaults to the last 30 days if omitted.
• get_session_funnel_journey — Get detailed session-level data with full funnel event sequences, behavioral tags, and checkout outcomes. Each row is one session with its complete journey: funnel events (PAGE_LOAD → PRODUCT_VIEWED → PRODUCT_ADDED → CHECKOUT_STARTED → ...), behavioral metrics (rage_click counts, dead_click counts, etc.), intention score, CRO analysis, and checkout result. Returns multiple sessions by default (up to limit). Use session_id to drill into one specific session, or filter by funnel/experiment/intention/behavioral_tag to find interesting sessions. Date range is optional — defaults to the last 30 days if omitted.
• get_session_behavioral_insights — Get aggregate behavioral patterns across session recordings. Shows how sessions break down by intention score, checkout completion, traffic source, and device type, with engagement metrics (avg duration, scroll depth, clicks). Use this to understand behavioral patterns like 'high-intent Meta mobile sessions have 3x the rage clicks'. Filter by behavioral tag (rage_click, dead_click, u_turn, excessive_scrolling, error_click, confusion_loop) to focus on specific UX problems. Date range is optional — defaults to the last 30 days if omitted.
• get_ad_creatives — Get ad creative metadata including ad copy, image URLs, video URLs, and video duration. Use this to understand what creative assets are running for each ad. Filter by ad_channel, campaign_id, or ad_set_id. Returns creative_id, ad_copy, ad_image_url, video_source_url, and video_length_seconds.
• get_ad_insights — Get per-ad performance metrics showing which individual ads, ad sets, and campaigns drive traffic and conversions to each destination. Shows ad spend, clicks, impressions, reach, in-unit sessions/purchases/revenue, ROAS, CPA, CVR, and AOV at the ad level. Filter by ad_channel (Meta/TikTok), campaign_id, ad_set_id, fermat_shop_id, or routing_config_id. Use this to identify top-performing and underperforming ads. Date range is optional — defaults to the last 30 days if omitted.
• get_product_variant_metrics — Get per-product-variant performance metrics within funnels from GA4. Shows which SKUs are driving sessions, views, add-to-carts, purchases, and revenue within a specific funnel or experiment layout. Filter by product_funnel_id, product_funnel_layout_id, or experiment config. Date range is optional — defaults to the last 30 days if omitted.
• get_heatmap_scroll_analysis — Get AI-analyzed scroll depth data for a funnel. Returns scroll depth markers bucketed at 5% intervals showing how many visitors reached each depth, drop-off points, consistency score, and total visitors. Works for both DPP funnels and AIS funnels. Scope to a specific shop, tile, product, bundle, product_group, product_funnel, or product_funnel_layout using optional ID filters. Date range is optional — defaults to the last 30 days if omitted.
• get_heatmap_click_analysis — Get AI-analyzed click heatmap data for a funnel. Returns clustered click areas with labels (e.g. 'Add to Cart button'), click percentages, confidence scores, top engagement areas, and conversion rate. Works for both DPP funnels and AIS funnels. Scope to a specific shop, tile, product, bundle, product_group, product_funnel, or product_funnel_layout using optional ID filters. Date range is optional — defaults to the last 30 days if omitted.
• get_destination_insights — Get destination-level performance insights combining ad spend, GA4 conversions, halo attribution, and in-unit metrics. Shows ROAS, CPA, CVR, revenue, purchases, ad spend, and halo allocations per destination broken down by ad channel. This is the primary tool for ad performance data — use it to answer questions about ROAS, CPA, ad spend, impressions, clicks, and cost efficiency across Meta, TikTok, Google Ads, and other channels. Date range is optional — defaults to the last 30 days if omitted.
• get_experiment_results — Get performance results for any experiment returned by list_experiments. Automatically routes to the correct data source based on experiment_type: shop_experiment types return layout variant comparison (sessions, conversions, revenue, CVR, AOV per layout); routing config types return Fermat vs external comparison (ROAS, CPA, revenue, CVR, purchases, traffic, spend, significance); product_group_experiment types return per-layout-variant DPP metrics (sessions, PDP views, add-to-carts, checkouts, purchases, revenue, CVR, revenue/session, revenue/order) from GA4 or Fermat (Claire) attribution — plus statistical significance analysis when an External variant is present: Revenue/Session on source=fermat (Claire), CVR/ATC Rate/Checkout Rate on source=ga4. Stat sig includes lift, confidence, Bayesian probability, confidence intervals, and Holm-Bonferroni correction for multi-layout experiments. attribution_mode selects in-session (default) or 1-day-click attribution (fermat source only). Use the experiment_type and relevant ID from list_experiments output. Date range is optional — defaults to the last 30 days if omitted.
• list_experiments — List all experiments for an organization — covers internal layout A/B tests (shop_experiment), URL routing split tests (weighted_routing_config, geo_routing_config, user_agent_routing_config), and product funnel experiments (product_group_experiment — A/B tests created within Dynamic Product Pages under a Product Group). Returns experiment metadata, status, dates, and related details. Filter by status or experiment_type, or omit filters to get everything.
• get_funnel_metrics — Get funnel performance metrics including sessions, conversions, revenue, CVR, and AOV. Data comes from the analytics pipeline (BigQuery → dbt → Cube.js). Filter by funnel_id and date range. Group by day/week/month. Date range is optional — defaults to the last 30 days if omitted.
• get_sov_report — Check the status and retrieve results of a Share of Voice (SoV) report. Use this to poll after starting a report with generate_sov_report. Status transitions: pending → processing (with progress 0-100) → completed. When completed, returns prompts (per-prompt SoV data) and unstable (aggregate stats).
• get_ais_crawler_analytics — Get AI crawler activity on an org's funnels. Shows which AI platforms (OpenAI, Anthropic, Perplexity, Google AI, etc.) are crawling each funnel, total bot visits, and per-crawler breakdown. Higher crawl frequency correlates with better AI search visibility. Optionally filter by start_date and end_date.
• get_ais_citations — Get AI Search citation data showing how many times an org's funnels appear in AI search results (ChatGPT, Perplexity, Google AI, etc.). Returns total citation count, per-funnel citation breakdown, and individual citation details including the prompt/keyword, search source, rank, and matched funnel. Requires start_date and end_date. Optionally filter by probe_type or search_source.
• get_ais_prompts — Get AI Search prompt contents for an org. Optionally filter by a specific prompt key.
• list_forever_links — List forever link configs and their routing destinations for an org. Shows link name/slug, current routing config with destinations (Fermat shops or external URLs) and traffic weights. Shows the most recent routing config per link with active destinations. Supports pagination.
• get_destination_details — Get full details of a single product funnel (destination) by its ID, including product groups, layouts, and live product counts. Returns product groups with their layouts, precedence order, and live product counts.
• list_funnels — List all funnels (shops) for an organization. Returns every funnel type: product funnels (DPP), AI search funnels (AEO/SEO), and regular shops. Each result includes tags, product funnel data (if DPP), AI search metadata (if AIS), and forever link routing info. Use the optional type filter to scope to a specific funnel type, or omit it to get everything. Supports search and pagination.

About MCP

The Model Context Protocol (MCP) is an open protocol Anthropic introduced for connecting LLM-based agents to external tools and data sources. Providers publish MCP servers that expose their API surface as structured, discoverable tools — an MCP-compatible client (Claude Desktop, Cursor, Cline, Continue, etc.) can connect to the server and call its tools without any per-provider integration code.

Browse every MCP server on the APIs.io network or compare with the broader Agent Skill surfaces of the same providers.