MotiBlog

One-call project digest: article counts by lifecycle status, the REVIEW queue with word counts, posts published in the last 7 days, content plans due in the next 7 days, the latest published post, quota usage, the project's current autonomy level (L1 = nothing ships without agent review; L2 = clean articles auto-publish), the agent-memory index with note counts and recently updated titles, and — when Search Console is connected — which published posts Google has not indexed yet.

List the MotiBlog projects permitted by the current credential. A project-restricted OAuth grant returns only the projects the human selected plus projects this agent created; an account-wide grant returns every project owned by the account. Includes settings that matter to agents and quota usage. Project credentials are never returned.

Takes no parameters.

Create a new blog project for the authenticated account. The website must be reachable and the account must have an available project slot. A restricted OAuth grant automatically receives access to the project it creates. MotiBlog starts the initial analysis and content pipeline after creation. Setup and endpoint behavior: https://motiblog.ai/mcp.

Get one MotiBlog project in detail: pipeline settings (language, internalLinks, banner), positioning inputs (businessProfile, productTruth, manualPositioning) and quota state.

Update operational settings for an existing project: name/status, approval and auto-publish policy, writing model/language, fact checking, positioning, internal links, self-hosting, and banner configuration. Omitted fields stay unchanged.

Start the autonomous pipeline for a project: site crawl and analysis, topic selection, and article generation for scheduled plans. Returns immediately while processing continues asynchronously.

Get project setup/site-analysis pipeline state: whether a crawl/analysis run is active, its steps, and recent run history. This response does not include article-generation state or per-article phase telemetry.

Retry a failed pipeline run by id. The API starts a fresh run for the same owned project and returns immediately.

Suggest one specific, keyword-oriented blog topic and add it to the project content plan as a DRAFT entry. Draft plan entries do not trigger article generation.

List the content plan queue for a project with each entry status: DRAFT (proposed), APPROVED (cleared for generation), IN_PROGRESS, COMPLETED.

Approve a DRAFT content-plan entry so it becomes eligible for article generation by an on-demand or scheduled pipeline.

Rewrite a content-plan entry with AI, producing a new title, summary, and target keyword. Available only before an article has been generated from the entry.

Move a content-plan entry to a new ISO 8601 scheduled date.

Add a fully authored content-plan entry with an exact title, target keyword, summary, content type, funnel stage, and date; the supplied fields are not AI-reinterpreted. Without scheduled_date, the entry is placed after the latest active plan. status=APPROVED makes it eligible for the daily scheduler; the default DRAFT status does not.

Edit the title, target keyword, summary, content type, funnel stage, scheduled date, or status of a DRAFT or APPROVED plan entry. DRAFT pauses scheduler eligibility without deleting the entry. IN_PROGRESS and COMPLETED entries are immutable records of generating or completed work.

Move up to 100 plan entries to new dates in one transactional call — reorder a month of the calendar without N sequential reschedules. Every entry must name an active plan in the project; any miss fails the whole batch, so a partial reorder never ships.

Soft-delete an ungenerated DRAFT or APPROVED content-plan entry, removing it from the plan list, calendar, digest, and daily scheduler while retaining it for restoration. SKIPPED or FAILED attempts do not block deletion; generating and completed entries cannot be deleted.

Restore a soft-deleted content-plan entry to the backlog with its prior status and scheduled date intact.

List soft-deleted content-plan entries for a project, most recently deleted first, including the ids and state needed for audit or restoration.

Generate an article from an APPROVED content-plan entry through research, outline, draft, fact-check, polish, and scoring phases. Returns the article in GENERATING state while processing continues asynchronously. When the article allowance is used up it returns a structured article_allowance_exhausted error.

Get scheduled content-plan entries (with slim article previews) between two ISO dates, e.g. start_date=2026-09-01 end_date=2026-09-30.

List articles awaiting attention, filtered by status. Lifecycle: DRAFT → GENERATING → REVIEW → APPROVED → PUBLISHING → PUBLISHED (also FAILED, SKIPPED). Default status=REVIEW (needs human/agent approval); pass APPROVED to see what is cleared for publishing.

Get full article markdown plus an always-present qualityGate summary of the latest content, link, and final-review blockers. Also returns factCheckReport, topicGate, seoScore, and publishedUrl; SEO score never overrides blockers. include_logs=true appends raw phase telemetry.

Edit an article by replacing markdown content, changing the title, metaDescription, or article-specific banner prompt, or moving its lifecycle status. Content edits can resolve fact-check flags before an article is moved to APPROVED or returned to DRAFT.

Approve an article through the governed publication gate. publish_now=true also triggers immediate publishing to enabled integrations or the self-hosted blog. Otherwise the article remains APPROVED for a scheduler, auto-publish policy, or later targeted publication.

Set or clear a per-article scheduled publish time. Pass exactly one of scheduled_for (an ISO datetime) or clear_schedule=true. Scheduled APPROVED articles are published automatically by the 15-minute scheduler when autoPublish is on.

Wipe an article and rerun its entire research and generation pipeline from scratch. This replaces the full article rather than editing a passage or one section.

Regenerate one chapter identified by a zero-based chapter_index while leaving all other article sections byte-identical.

Per-phase pipeline telemetry for one article: phase names, statuses, messages, token counts, and USD cost per phase. Diagnose FAILED articles or audit what the generator did.

Aggregate reader feedback for one article: rating counts and average from the public rating widget, without individual comments.

List content that may benefit from a refresh across all projects permitted by the current credential. source="article" identifies published articles with Search Console decay signals; source="external_page" identifies pre-existing customer CMS pages flagged after a near-duplicate draft, with the rationale in reason.

Takes no parameters.

Dismiss an active refresh recommendation. article_id accepts a published article suggestion with source="article" or a pre-existing crawled-page suggestion with source="external_page".

Bulk-generate up to 500 articles from structured rows and a title template such as "Best {{product}} for {{location}}". Creates approved plans and enqueues generation; the same article usage limits apply.

Rewrite a selected passage with an inline AI action: improve, shorter, longer, grammar, active-voice, or add-example. Returns replacement text without modifying any stored article.

Download a JPEG, PNG, or WebP from a public HTTPS URL (maximum 5MB), upload it to MotiBlog, and set it as an article banner. Replaces the current banner.

Remove the current banner image from an article.

Download a JPEG, PNG, or WebP from a public HTTPS URL (maximum 10MB) into MotiBlog storage and return its permanent URL for article content, author avatars, or banner references.

Get the saved banner-style preview, prompt, and remaining lifetime preview allowance for a project.

Suggest a short visual prompt for a banner preview using the project niche and optional style preset. This does not generate an image or consume preview allowance.

List configured publishing targets (integrations) for a project: WEBHOOK, WORDPRESS, GHOST, WEBFLOW, SHOPIFY, DEVTO, SANITY, CUSTOM_API — with enabled state.

Get one publishing integration and its enabled/last-used state. Stored provider credentials and config values are redacted from tool output.

Create any supported publishing target: WEBHOOK, WORDPRESS, WEBFLOW, CUSTOM_API, GHOST, SHOPIFY, DEVTO, or SANITY. Pass the provider-specific config object; credentials are accepted but never echoed by MCP read tools.

Edit a publishing target name, type, provider config/credentials, or enabled state. Omitted fields remain unchanged.

Permanently delete a publishing integration. Existing publish logs remain, but this target can no longer receive content.

Register a webhook publishing target: MotiBlog POSTs the published article JSON to your endpoint (signed with X-Signature: sha256=<HMAC(secret)> when secret is set). Config shape: { url, secret?, method? }. This is how an agent's own infrastructure receives content pushes. Setup and endpoint behavior: https://motiblog.ai/mcp.

Test a publishing target connection. mode=ping (default) checks credentials/reachability only; mode=full creates a real draft test item on the target and does not delete it automatically.

Push an APPROVED article to one specific integration by enqueueing a publish job that transitions from APPROVED to PUBLISHING and then PUBLISHED. Publishing history records the result.

Retry a failed publish attempt for an article/integration pair.

Mark an article and integration pair as manually synced, creating an auditable synthetic success log for content that already exists at the target.

Publishing history: SUCCESS/FAILED attempts with publishedUrl, provider error text and timestamps. Self-hosted publications are represented as synthetic SELF_HOSTED success records because they do not use an integration. Pass article_id for one article, or omit for the project.

List tracked keywords for a project (search volume, difficulty, intent, status). Keywords drive topic relevance in generation.

Track a new SEO keyword for a project so future topic suggestions/generation can target it. intent is one of INFORMATIONAL, NAVIGATIONAL, TRANSACTIONAL, COMMERCIAL.

Change a tracked keyword lifecycle status: DISCOVERED, APPROVED, REJECTED, or USED.

Permanently remove a tracked keyword from the project.

List the competitor domains tracked for a project, including overlap and domain-authority signals when available.

Add a competitor domain to a project so competitive analysis and content-gap planning can use it. Setup and endpoint behavior: https://motiblog.ai/mcp.

Permanently remove a competitor from the project.

List pillar clusters with their pillar keyword, member keywords, content plans, article status, published URLs, and completion percentage.

List the project proprietary facts — ground-truth statements about the product (features, pricing, integrations) injected into prompts and enforced by strict fact-checking.

Get one proprietary product fact by id.

Supply a ground-truth fact about the customer's product, or update one via fact_id. Strict fact-checking uses facts available before generation to verify claims. Example: fact='Acme integrates natively with Shopify since v2.3', category=['integrations'].

Permanently remove a proprietary product fact from generation context.

Index of this project's persistent agent notes across sessions and agent vendors. Returns key, title, kind, pinned flag, a short summary, and updatedAt for every note without returning note bodies.

Read one agent note in full (markdown body) by its key.

Create or update a persistent project note by key for strategy, research findings, decisions and their reasons, or rejected approaches. Existing keys are updated instead of duplicated. pinned=true includes up to the five newest pinned notes, with dates, in content-planning context; dashboard settings take precedence on conflict. Limits: 200 notes per project and 50,000 characters per body. Setup and endpoint behavior: https://motiblog.ai/mcp.

Permanently delete one agent note by key. This is intended for notes that are incorrect or no longer needed.

Semantic search over a project's agent notes. Returns keys, titles, and 300-character excerpts ranked by similarity. Recently saved notes may not be searchable immediately because embedding is asynchronous.

Get the computed brand-voice profile for a project: tone, style, vocabulary, cadence, and extraction metrics.

Preview the exact brand-voice context that article generation would receive for a target keyword.

Re-extract the project brand voice asynchronously from current source content. The updated profile becomes available after processing completes.

List every author profile available to a project.

Get one author profile with bio, avatar URL, and social links.

Create an author profile for article attribution.

Edit an author name, bio, avatar URL, or social links.

Permanently delete an author profile from the project.

Get the sitemap, blog-root, or manual URL source used to discover pages for internal-link intelligence.

Set the discovery source for content intelligence: SITEMAP, BLOG_ROOT, or MANUAL, plus its URL. Setup and endpoint behavior: https://motiblog.ai/mcp.

Discover or refresh pages from the configured sitemap/blog source for internal-link intelligence.

List pages discovered for content intelligence, optionally filtered by text search.

Remove one discovered page from content intelligence.

Get content-intelligence page totals, classification counts, and embedding state.

Start website crawling. mode=initial runs the full initial analysis; recrawl refreshes pages without LLM analysis; full_reanalysis refreshes pages and reruns AI analysis.

Get current crawl progress: phase, pages crawled, and total pages.

List crawled website pages without their large full-content fields.

Get one crawled page including its full extracted markdown content and link metadata.

Get site analysis: ICP, positioning, competitors, content gaps, opportunities, and keyword clusters.

Get Search Console connection, selected-property, and latest sync state for a project.

Start a Google Search Console connection and return an authUrl for the human account owner to complete Google consent.

List Google Search Console properties accessible through the connected Google account.

Select which connected Search Console property belongs to this MotiBlog project.

Disconnect Google Search Console from the project and remove its stored credentials.

Get the earliest and latest Search Console dates stored for this project.

Get Search Console performance totals and time series over an optional YYYY-MM-DD date range (defaults to the last 28 days).

List Search Console queries with clicks, impressions, CTR, and position. Supports date range, sorting, pagination, and text search.

List Search Console pages with clicks, impressions, CTR, and position. Supports date range, sorting, pagination, and text search.

List Search Console insights, optionally filtered by insight type and status.

Dismiss a Search Console insight so it leaves the active recommendation queue.

Get up to 90 days of Search Console ranking history for a published article.

Get the agent action queue derived from Search Console: ranking, rewrite, backlink, and not-indexed recommendations tied to plans/articles.

Start a manual Search Console data sync. Returns immediately while refreshed performance, page, and query data are collected asynchronously.

Get the Google indexing state of every published article from stored URL Inspection results: how many are confirmed indexed and which are published but not (yet) indexed, with coverage state per URL. No Google API calls — reads the daily indexing check.

Live-check 1-10 URLs against Google's URL Inspection API: indexing verdict, coverage state, last crawl time, and Google's chosen canonical. Requires a connected Search Console property that covers the URLs. Results are cached onto matching published articles.

Get Google Analytics 4 connection, selected-property, and latest sync state for a project.

Read aggregated GA4 landing-page sessions, active users, key events, and revenue already synced for a project. Returns totals and the highest-session pages for a 1–90 day window.

Start a Google Analytics 4 connection and return an authUrl for the human account owner to complete Google consent. GA4 uses its own consent (analytics.readonly), separate from Search Console.

List GA4 properties accessible through the connected Google Analytics account.

Select which connected GA4 property belongs to this MotiBlog project and start the initial data sync.

Disconnect Google Analytics 4 from the project and remove its stored credentials and synced data.

Start a manual GA4 data sync. Returns immediately while per-page sessions, active users, key events, and revenue are collected asynchronously.

Join search demand (Search Console impressions, clicks, position) with outcomes (GA4 sessions, key events, revenue) per page over the last 28 days, classified into action buckets: CONVERSION_MISMATCH (traffic converts nothing — rewrite for intent), CTR_FIX (demand but weak titles), SCALE_WINNER (converts from page 2 — push it), STRIKING_DISTANCE (position 11-20), DEAD (no demand, no visits). Each row carries a recommendation. Works with Search Console alone; connecting GA4 unlocks the conversion buckets.

Get the account's article usage: articles used, the article limit, articles remaining, project slots, and when the usage period ends. Read-only.

Takes no parameters.

Get the authenticated MotiBlog account identity: email, display name, avatar, and verification state. Internal account and sign-in-provider identifiers are never returned.

Takes no parameters.

Update the authenticated account display name and/or avatar URL. This cannot change email, password, or OAuth grants.

List recent account notifications, optionally only unread ones.

Get the number of unread notifications for the authenticated account.

Takes no parameters.

Mark one account notification as read.

Mark every unread account notification as read.

Takes no parameters.

In local stdio MCP only, export articles as deployment-ready files for Astro, Next.js, Jekyll, Hugo, or git codebases. Remote HTTP MCP cannot write to the caller filesystem and returns a clear remote_filesystem_unavailable error instead of a server-local path.