Lantern MCP connector

Let an AI assistant or app that asked for access to your account use it. Use get_notifications to find waiting requests. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.approve_binding_request

Set which brands a teammate can open. This replaces their current list, so any brand you leave out is taken away from them. An empty list is the exception: it gives them every brand. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.set_member_brands

Claim a website Lantern already knows belongs to you, for example one found when you connected your store. No DNS record is needed. For any other website, use verify_domain. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.adopt_domain

Get a link that connects Slack to your account. Someone has to open the link and approve it in Slack; the assistant cannot finish this step. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.connect_slack

Turn down an AI assistant's or app's request for access to your account. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.reject_binding_request

Disconnect a store from your account and remove Lantern's access to it. Reconnecting means signing in to the store again. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.disconnect_store

End a session in which an assistant is working for this brand. The session stops at once and cannot be restarted.

Name your assistant sees: agent.terminate_delegated_session

Get a summary of your Lantern account: its name, plan, and how many members and brands it has.

Name your assistant sees: account-manager.get_account

Check whether Google Analytics is connected, and list the properties Lantern reads.

Name your assistant sees: account-manager.get_ai_traffic_connections

Give Lantern writing that shows how the brand sounds: pasted text, an interview transcript, or the address of one page, such as a style guide or About page. A page is read right away; if it cannot be read, the source is still saved and the result says the text could not be pulled, so do not add it again. To upload a file, use the brand voice settings in Lantern. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.add_voice_source

Add pages Lantern has already read from the brand's website as brand voice sources. Each run adds the next batch, and running it again never adds the same page twice.

Name your assistant sees: account-manager.seed_website_sources

Answer the current brand voice interview question, point to a source you already added, or both. The result is the next question, or says the interview is done.

Name your assistant sees: account-manager.reply_voice_interview

Build a brand voice from the sources added so far. It runs in the background for a few minutes and ends as a proposed version for you to review; it never goes live by itself. Your credit balance is checked first. This does not take credits from it, but a voice that finishes building counts 5 credits toward the brand's monthly agent credit cap, if one is set. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.synthesize_brand_voice

Replace the live brand voice with a full voice document you supply. Lantern checks it first; if anything is wrong, nothing changes and the result points to the exact problem. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.update_brand_voice

End the brand voice interview and save everything answered so far as a source, even if some questions were skipped.

Name your assistant sees: account-manager.finish_voice_interview

Get the brand's current voice (tone, audience, phrases to avoid, key messages) and any new version waiting for review.

Name your assistant sees: account-manager.get_brand_voice

List the material the brand voice is built from (text, web pages, interviews, uploads) and whether each one has been processed.

Name your assistant sees: account-manager.list_voice_sources

See what Lantern can do through a store platform's connection, such as inviting staff or syncing products. Something missing from the list is unknown, not unavailable. The Lantern pixel works on every platform; use get_pixel_install_snippet for it.

Name your assistant sees: account-manager.inspect_platform_capabilities

Add a new brand to your account, with its name and at least one website. Your plan sets how many brands you can have; if you are at the limit, the result says so.

Name your assistant sees: account-manager.create_brand

Match a description like "my UK Shopify" to one of the connected stores. Returns the match, or a short list to choose from when more than one fits.

Name your assistant sees: account-manager.resolve_store_reference

Get one brand's name, short name and websites. Leave brandId out to use the brand selected for this connection.

Name your assistant sees: account-manager.get_brand

Check whether a website you claimed is verified, and get the DNS record that proves you own it.

Name your assistant sees: account-manager.get_domain_status

See how far the brand's setup has come: whether the first scans have finished, which setup steps are done, and anything blocking the rest.

Name your assistant sees: account-manager.get_onboarding_status

Show which theme is live on each connected store, when it was published, and its status. Pass store_id to check one store.

Name your assistant sees: account-manager.get_theme_status

Get the DNS record that proves you own the brand's website during setup. Add it as a TXT record, then run verify_domain. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.verify_onboarding_domain

For one of the brand's pages, see how often each AI assistant cited it, how prominently, and for which prompts, newest first.

Name your assistant sees: findings-interpreter.explain_citation_telemetry

List the brand's pages that AI assistants cite, with how often and how prominently each assistant cites them. Includes history imported from a previous provider when there is some. Filter by date and assistant.

Name your assistant sees: findings-interpreter.get_brand_citation_telemetry

List the pages where AI assistants cite competitors alongside the brand, most competitor citations first.

Name your assistant sees: findings-interpreter.get_brand_competitor_citations

Start tracking a competitor for the brand. A brand can track up to three. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.add_competitor

Add a starting list of competitors during setup, usually the ones AI answers already mention alongside the brand. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.seed_competitors

Start an AI readiness analysis of a competitor's website and wait for it to finish, with progress updates along the way. Pass the same idempotency_key again within a day to get the first run back.

Name your assistant sees: monitor.analyze_competitor

See how often AI assistants cite each competitor compared with the brand, on the prompts the brand tracks: citation counts, share of voice, typical position in the answer (lower is better), and which assistants cite them. This is relative to the brand's own prompts, not a market share.

Name your assistant sees: findings-interpreter.compare_brands

Confirm a competitor Lantern suggested, so it is tracked as one you chose. If you removed it earlier, add it again with add_competitor instead.

Name your assistant sees: account-manager.verify_competitor

See every brand in your account together with its confirmed competitors, in one call. Useful for spotting competitors your brands share.

Name your assistant sees: account-manager.get_brand_graph

List the competitors tracked for the brand. Only confirmed competitors are included unless you set includePending. Each competitor's id is what analyze_competitor takes.

Name your assistant sees: account-manager.get_competitors

Turn down a competitor Lantern suggested, so it stops appearing for this brand. Rejections also teach Lantern: once five different accounts reject the same pairing, Lantern stops suggesting it to anyone, and removes it where other accounts have not confirmed it themselves.

Name your assistant sees: account-manager.reject_competitor

Approve a draft that is waiting for review, so it is ready to publish. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: content-studio.approve_draft

Delete a draft for good. Only drafts that are waiting for review or approved can be discarded. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: content-studio.discard_draft

Hide a content idea from list_content_suggestions for a while. It takes a content idea's id, not an opportunity's; to skip an opportunity, use dismiss_opportunity.

Name your assistant sees: content-studio.dismiss_content_suggestion

Write a piece of content for the brand: an expert guide, product story, FAQ, help article, community reply or pitchable article. It is written in the background over a few minutes and lands in your drafts for review; it is never published by itself. To write for a content opportunity, pass its opportunityId. The result has a link to the draft. Your credit balance is checked first. This does not take credits from it, but each draft that finishes counts 3 credits toward the brand's monthly agent credit cap, if one is set. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: content-studio.draft_content

Get a draft's full content, what it was based on, and where it will be published.

Name your assistant sees: content-studio.get_draft

List up to five content ideas for the brand, each with a short reason, and which content opportunities each one answers.

Name your assistant sees: content-studio.list_content_suggestions

List the brand's content drafts. Filter by status to see, for example, only the ones waiting for review.

Name your assistant sees: content-studio.list_drafts

Record that you placed an approved draft yourself, for example on your own site or as a community reply. Lantern does not post anything. Only approved drafts can be marked.

Name your assistant sees: content-studio.mark_draft_published

Get everything about one suggested fix: what kind it is, the exact change, when it was created or applied, and whether it was verified. Read this before applying or explaining a fix.

Name your assistant sees: findings-interpreter.get_action_metadata

See how much a suggested fix is expected to help, how confident Lantern is in that, and how large a change it makes, so you can weigh it before applying.

Name your assistant sees: findings-interpreter.get_action_impact

Get before and after images of a page for one recommendation. The links expire after about 15 minutes, so call again for fresh ones. An empty list means no mockups have been made yet.

Name your assistant sees: findings-interpreter.get_mockup

List things worth writing about for the brand, each with a headline and the reason, best first.

Name your assistant sees: content-studio.list_content_opportunities

List up to 10 suggested fixes waiting for your review, newest first. For recommendations with expected impact, get_brand_overview gives a fuller picture.

Name your assistant sees: findings-interpreter.get_action_queue

Get one persona by its id or exact name, including how it is defined and its products, pages and questions. When a figure is held back, give the reason exactly as returned; ordersAreLowerBound means the real order count may be higher. The Personas page in Lantern shows the same figures.

Name your assistant sees: findings-interpreter.get_audience

List the brand's personas, the groups of shoppers who behave alike, with the visits and orders recorded for each. When a figure is held back, say it is held back, never zero, and give the reason exactly as returned. Revenue across mixed currencies is not available. The Personas page in Lantern shows the same figures.

Name your assistant sees: findings-interpreter.list_audiences

Start an AI readiness analysis for one product and wait for it to finish, with progress updates along the way. Pass the same idempotency_key again within a day to get the first run back.

Name your assistant sees: monitor.analyze_product

Get the number of products in the brand's catalog. Grades and scores are not filled in yet.

Name your assistant sees: findings-interpreter.get_product_catalog_summary

See how complete and clear one product's data is: its score, grade, the biggest issues and when it was last checked.

Name your assistant sees: findings-interpreter.get_product_analysis

List products from the brand's catalog. Filter by store, search text, vendor or product type, and sort by title or sync date. Scores and grades are not filled in yet.

Name your assistant sees: findings-interpreter.get_products

Bring in the latest products from a connected Shopify store. Your plan limits how many products Lantern keeps for each store; products past that limit are archived. The result counts what was created, updated and archived.

Name your assistant sees: monitor.sync_store

Add a tracked prompt to one of the brand's collections. Adding one that is already there changes nothing. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.add_prompt_to_collection

Hide one of Lantern's prompt suggestions so it stops appearing.

Name your assistant sees: account-manager.dismiss_suggestion

Change a tracked prompt: its wording, whether it runs, whether it is a favorite, or whether it counts as a branded prompt. Give only what you want to change. New wording that matches another prompt the brand tracks is refused.

Name your assistant sees: prompts.edit_prompt

Estimate how many credits a brand's tracking will use each month, with the prompt count, schedule and AI assistants behind the number. Nothing is charged. Share this before turning tracking on.

Name your assistant sees: account-manager.estimate_visibility_credits

Read what AI assistants actually answered for one tracked prompt. With no dates, you get the latest answer from each assistant; with start_date or end_date, every answer in that window, newest first. Identify the prompt by prompt_id or by its wording. Each answer is marked "answered" or "no_answer". "no_answer" means the assistant gave no answer (for google_ai_overview, Google showed no AI Overview for that search). It is not a miss: its text is empty, mentioned and rank are null, and it counts toward no visibility score.

Name your assistant sees: findings-interpreter.get_prompt_responses

Get the latest result for each tracked prompt. This currently returns an empty list; use get_prompt_responses for the AI answers to a prompt.

Name your assistant sees: account-manager.get_query_results

Ask Lantern for new prompt suggestions for the brand, researched the same way as its weekly suggestions. They appear in list_suggestions for you to track or dismiss. Your credit balance is checked first. This does not take credits from it, but each run that finds new suggestions counts 5 credits toward the brand's monthly agent credit cap, if one is set. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.propose_suggestion

See how tracking is set up for the brand: whether it is on, how often it runs, and how many prompts it tracks.

Name your assistant sees: account-manager.get_brand_visibility_settings

Check whether a report you asked for is ready. When it is, the result has its link; otherwise it says it is still being made. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.claim_report

Email a summary report. It goes to the account owner unless you give another address. Sent email cannot be recalled. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: account-manager.email_report

Estimate how many credits a report will cost and whether your balance covers it. Nothing is charged.

Name your assistant sees: account-manager.estimate_report_credits

Check whether a report is still being made, is a draft, is published or failed. Published reports include their share and PDF links.

Name your assistant sees: account-manager.get_report_status

List the reports made for the brand.

Name your assistant sees: account-manager.list_brand_reports

Make a recommended change live on your store, for example updated product copy. Lantern keeps a snapshot first, so rollback_change can undo it. Your credit balance is checked first. This does not take credits from it, but each change it makes live counts 5 credits toward the brand's monthly agent credit cap, if one is set. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: site-editor.apply_recommendation

Mark a recommendation as approved. This does not change your store; apply_recommendation does that.

Name your assistant sees: site-editor.approve_recommendation

Approve a store task so Lantern can carry it out. Approving does not carry it out by itself.

Name your assistant sees: site-editor.apply_action

Make an earlier llms.txt version your current text again. It is not live until you run prepare_llms_txt. Depending on the brand's approval settings, the change may wait for someone to approve it in Lantern; the result then says it is proposed.

Name your assistant sees: site-editor.restore_llms_txt_version

Choose whether a store task edits the item that is live now or creates a new one.

Name your assistant sees: site-editor.set_action_strategy

Reserve a store task for this conversation so no other assistant picks it up at the same time.

Name your assistant sees: site-editor.claim_action

Connect the service that runs your website's DNS, so Lantern can manage the records it needs.

Name your assistant sees: site-editor.connect_dns_provider

Decline a store task so it leaves the list of ones waiting for you.

Name your assistant sees: site-editor.refuse_action

Start an AI readiness analysis of the brand's website and wait for it to finish, with progress updates along the way. Pass the same idempotency_key again within a day to get the first run back instead of starting another.

Name your assistant sees: monitor.analyze_domain

Get a summary of every brand in the account: each one's status, fixes waiting for review and whether its first analysis has finished. Use this to compare brands.

Name your assistant sees: findings-interpreter.get_portfolio_overview

Start here. Get the brand's latest scores, top recommendations and analysis status in one call. detail "summary" (the default) answers most questions; "full" adds every recommendation and a page breakdown; "with_history" adds the last 12 weeks of scores; "with_per_query" adds results per prompt, including which competitors were cited. Only go past "summary" when asked for that detail. Describe recommendations by their title and explanation, never by score or rank.

Name your assistant sees: findings-interpreter.get_brand_overview

Get a quick summary of the brand: its status, connected websites, fixes waiting for review, and whether the first analysis has finished. Use this when asked "how am I doing?"

Name your assistant sees: findings-interpreter.get_overview

See the words and phrases AI assistants use to describe the brand, whether each one is positive or negative, and how often it came up compared with the period before.

Name your assistant sees: account-manager.get_descriptor_trends

See how visible the brand is in AI answers over time: the latest score and a score per day for each AI assistant. For the full picture, start with get_brand_overview.

Name your assistant sees: findings-interpreter.get_brand_scores

List what needs attention for the brand: fixes waiting for review or that failed, disconnected websites, and setup that has not finished. Use this when asked "what should I do next?"

Name your assistant sees: findings-interpreter.get_overview_attention

Run a tracked prompt now on the AI assistants this brand tracks, and save the answers. Each run uses credits, the same as a scheduled run, and counts toward the plan's monthly runs. If the account is out of credits or has reached a limit, the prompt does not run and the result says why. Only active prompts can run.

Name your assistant sees: monitor.run_prompt

Fetch and return the content of a web page by URL. Useful for reading a specific page referenced in a recommendation or cited in a prompt. Pass-through to the Commerce Engine `engine.fetch_page` tool; the engine handles SSRF protection and content sandboxing. Required scope: read:findings.

Name your assistant sees: findings-interpreter.fetch_page

Fetch Google Analytics 4 AI-referral traffic insights for the brand: total sessions and conversions attributed to AI referral sources (ChatGPT, Perplexity, Claude, Gemini, etc.) over the last 30 days, broken down by provider. Requires the brand to have connected Google Analytics 4. Read-only; requiredScope: read:findings.

Name your assistant sees: findings-interpreter.get_ai_traffic_insights

Fetch Google Search Console insights for the brand domain: crawl stats (total requests, AI agent hits, response-code breakdown), top 10 organic queries (clicks, impressions, CTR, position), and estimated indexed page count. Requires the brand to have connected Google Search Console. Read-only; requiredScope: read:findings.

Name your assistant sees: findings-interpreter.get_search_console_insights

Search the web for current information relevant to the brand or a user question. Returns a list of web results with titles, URLs, and snippets. Pass-through to the Commerce Engine `engine.web_search` tool; the engine handles SSRF protection, rate limiting, and content sandboxing. Required scope: read:findings.

Name your assistant sees: findings-interpreter.web_search