uplika MCP server

• approve_replyFree

  Threads only: approve (or ignore) a reply held by reply approval on one of the account's posts. Approving makes it public. Read the queue with pending: true.

• bridge_statusFree

  Naver Blog only. Posts to Naver Blog are written by the uplika browser extension inside the user's own Chrome, so nothing goes out while that Chrome is closed. Call this before publishing to Naver Blog. If online is false, the response carries wake commands per OS that open Chrome on the user's computer in the profile that has the extension (found by extension id). If you have a shell on that computer, run the one for its OS, then call this again or get_post; the extension reconnects within about a minute and queued posts go out. publish also returns the same bridge object when the extension is offline. state is one of online, offline, logged_out (Chrome is on but not logged in to Naver). userMessage is a sentence in the person's language to relay as it is. queued is how many posts wait for the extension; delete_post cancels them and update_post rewrites them before they go out. extensionVersion and kinds say what that extension can do.

• create_automationwrite actionFree

  Create an automation from any template in list_automation_templates by passing its params. Creates a draft. For a flow no template covers, build the document yourself and call put_automation on the draft.

• create_comment_to_dmwrite actionFree

  The most common automation: when someone comments on a post, DM them. **Creates a draft; nothing goes out until enable_automation.** How Meta works: a DM cannot be started by the account. The only automatic door is a private reply to a comment, one per comment, within 7 days of the comment, and once the person answers the 24-hour window opens for the rest. Instagram can check whether the person follows the account (requireFollow); Facebook cannot, so requireFollow is rejected there. Threads has no DMs at all; use create_automation with comment_public_reply. Files go through media_presign first and are passed as deliver.mediaId. post can be our post id, the post's own id on the platform, a link to the post, "any" for every post, or "next" for the next post you publish (or pass automation on publish to do both in one call). Only one enabled automation per account can wait for "next" (409 next_post_taken). Order when everything is on: opening DM (message + button) -> askEmail -> requireFollow -> deliver (text, up to three link buttons, file; clicks are tracked) -> followUp if no link was clicked. openingDm:false sends deliver as the private reply itself; then requireFollow, askEmail, followUp and files are rejected because the window never opens.

• delete_postwrite actionFree

  Delete a post from the channel for good. This is not reversible, so confirm with the person first. Daily delete limits differ per channel: Threads 100 a day, Instagram only on accounts connected via Facebook, with no documented daily cap, YouTube 20 a day, Facebook 50 a day, Bluesky 35000 a day, Telegram only within 48 hours of publishing, with no documented daily cap, Naver Blog through the browser extension, with no documented daily cap, TikTok has no delete API; posts can only be removed in the app. On a scheduled or draft post nothing is on any channel yet, so this simply cancels it and removes our record. A Naver draft (target externalId starting with draft:) is only in Naver's draft box: this cancels our record and, with extension 0.7.2 or later, removes that draft too. On Threads and YouTube this also works on posts written in the channel's own app, given the link. Instagram only lets us delete on accounts connected via Facebook: an account connected with Instagram login cannot be deleted through us at all, so tell the person to delete it in the Instagram app. Facebook only lets us delete Page posts this app published, so a post made in the Facebook app cannot be deleted through us. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• delete_replywrite actionFree

  Delete a comment for good. This is not hide_reply: it cannot be undone. What it reaches differs by channel and list_platforms says which ones support it at all. On Instagram and Facebook it removes anyone's comment on your post; on Threads and Bluesky a reply is itself a post, so it only removes replies the connected account wrote. Prefer hide_reply when the person just wants it out of sight.

• describe_grammarFree

  How to write the body for a channel that has its own markup. Naver Blog has one; every other platform returns not_supported, which is not an error to work around. Call this before writing a Naver Blog body for the first time, or whenever you want something the basics do not cover: highlighting a phrase, a styled table, a collage, an event block, a map with several places. Without a topic you get an overview and the list of topics; with one you get that section in full, including the mistakes that fail silently. The values come from the same grammar the publisher validates against, so what this returns is what publish accepts.

• diagnose_naver_blogFree

  Naver Blog only. Reads a blog's public RSS feed (its latest posts, at most 50) and says whether the titles are written for search: the share of titles carrying a search intent word (price, how to, review), how many start with a date or episode label, posts per month, the categories, and the words the blog repeats in titles (seedCandidates: candidates to research, not proven keywords). summary.oldest and summary.newest say which dates it read, so a quiet blog's 50 posts are not everything; summary.capped is true at 50. flags and thresholds carry the judgement (lowIntent when the intent share is under thresholds.lowIntentPct). Works for any public blog, not only connected ones, and needs no Naver keys. Cached 24 hours; a cache miss spends one of 20 diagnoses per person per day (quota in the response). Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• disable_automationFree

  Stop an automation. Runs already waiting for a button stay waiting but nothing new starts.

• enable_automationFree

  Turn an automation live. **This is the moment messages start going to real people.** Confirm with the person first. Only one automation per account can wait for the next post: enabling a second one is refused with next_post_taken until the first binds to a post. Refuses with reconsent_required if the account was connected before DM permissions, and warns if the account is not subscribed to webhooks.

• expand_naver_keywordsFree

  Naver Blog only. Expands seed keywords one level through Naver autocomplete on the server and returns every word found with the seed it came from (found[].from is seed or L1:<seed>). These are candidates to measure with research_naver_keywords, not proven keywords. Up to 10 seeds. Cached seven days; a call with at least one uncached seed spends one of 20 expansions per person per day. Server-side autocomplete is off by default: when the server answers naver_autocomplete_unavailable, pass your own keyword list to research_naver_keywords instead. naver_autocomplete_paused and naver_autocomplete_busy carry retryAfterSeconds. Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• followFree

  Follow an account on the channel. Naver Blog only today: adds the blog as a neighbor. mutual: true sends a mutual-neighbor request that the other blog has to accept, so the result is pending until they do; without it the blog is added as a plain neighbor right away. Already a neighbor comes back as already. There is no unfollow. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• get_automationFree

  One automation as a document: triggers, nodes, start. Also returns version, which put_automation needs, and templateParams when the flow still has its template shape.

• get_contactFree

  One contact: name, username, whether they follow the account and whether it follows them (Instagram only, and only for people who have messaged), follower count, tags, opt-out, and when the messaging window closes. Follower lists do not exist on any channel; this is the closest thing.

• get_insightsFree

  Views, likes, replies, reposts, quotes and shares for one publish. Views and shares can be null when the platform does not report them yet — null is not zero. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• get_naver_keyword_historyFree

  The measurement history of one keyword: one point per day it was actually measured, newest first, with search volume, document count and posts per month. Shows whether a keyword is rising or cooling. An empty list means nobody has measured it here yet. Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• get_postwrite actionFree

  One publish: per-target status and the reason any target failed. For a post link you probably want open_post instead. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• get_publish_optionswrite actionFree

  What a channel needs to know before you publish to it. Only TikTok has this today; every other channel returns not_supported, which is not an error to work around. For TikTok it returns the creator nickname the post will go out as, the privacy levels this account may use right now, whether it can post at all, and its video length limit. Call it before every TikTok publish: the values are per account and change when the person edits their TikTok settings. options.tiktok.privacyLevel is required and has no default, so this is where you get the value to pass.

• get_quotaFree

  How much of the 24 hour allowance is already used for posts, replies and deletes. Check this before a burst of publishing. This is live usage from the platform, not the static limits in list_platforms. We also cap how fast one account can publish, so publish can return rate_limited even when the platform allowance still has room.

• hide_replyFree

  Hide a reply on the channel, or show it again with hide: false. The reply id comes from list_replies, and postId is the publish it belongs to.

• likeFree

  Like a post — or a comment: pass replyTo (a reply id from list_replies) to like that comment instead of the post. Idempotent: if it is already liked the call succeeds with already: true and nothing is toggled. There is no unlike. Works on Naver Blog and on Instagram accounts connected through Facebook; other channels answer not_supported. On Naver Blog the post can belong to another blog: pass its link and we act as the connected account. Naver ignores liking your own post, and that comes back as naver_not_allowed rather than a fake success. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• list_accountsFree

  Connected social accounts. **Call this before publishing anything.** Each item has id, platform (threads etc.), handle and status. The accountIds you pass to publish are these ids, and only "active" ones publish. If the person did not name a channel, target every active account. If the list is empty, no channel is connected yet. Send the person to https://uplika.com/dashboard/connections.

• list_automation_runswrite actionFree

  Runs of one automation with their status and a step log. Status names why a run stopped: blocked_window (24-hour window closed), blocked_opt_out, blocked_paused (a person is handling that conversation), blocked_burst, failed_channel, expired, superseded, waiting, done.

• list_automation_templatesFree

  Ready-made automation templates with the params each one takes. Read this before create_automation. Every template lists the channels it works on. The most used one is comment_to_dm: a comment on a post gets one private reply with a button, and tapping it delivers a link or file inside the messaging window.

• list_automationsFree

  Automations on the connected accounts: name, channel, whether it is live or a draft, triggers, run count. Use get_automation to read one flow's document.

• list_channel_postsFree

  What is actually on the channel right now, including posts written in the channel's own app. Use this to find a post when you do not have its link. Each item carries a permalink you can pass straight to open_post, reply or delete_post. An empty list does not always mean the account has no posts: on TikTok this reads public videos only, so it stays empty while the app is awaiting TikTok's Content Posting audit and every post goes out private. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• list_conversationsFree

  The inbox: DM conversations on Instagram and Facebook and mention threads on Threads. Each item says whether the 24-hour window is open and how long is left. state: open (default) or closed. kind: dm or mention.

• list_mentionsFree

  Threads posts that mention the connected account, delivered by webhook. Same shape as list_conversations with kind mention. Reply with send_dm, which posts publicly.

• list_naver_draftsFree

  Naver Blog only. Lists the drafts (temp-saved posts) sitting in the blog's draft box: logNo, title and the last-saved time, newest first. A post you published with options.naver_blog.draftOnly is one of them, and so is anything the person saved by hand in the Naver editor. Pass a logNo from here to publish_naver_draft. Needs the extension 0.7.2 or later in the person's Chrome; otherwise you get extension_outdated. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• list_naver_keyword_reportsFree

  Lists the person's saved keyword research (from research_naver_keywords or the dashboard), newest first: when, what they typed, the first set's main and sub keywords, and every set with its prompt. Reports belong to the person, not a workspace. Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• list_platformsFree

  Every channel and its rules: character limit, whether media is required, image and video limits, and what state the channel is in. Read this instead of guessing a platform's limits. status says who can connect: live means anyone; beta means the channel is in platform review and only accounts registered as testers on our app can connect yet, though publishing works normally for those accounts; bridge means it needs the user's browser extension running; soon means it is not connectable at all. charCount tells you how that channel counts a character, so you can check the length before calling publish instead of after it fails. options is the JSON Schema of what options.<channel> takes on publish: which fields exist, which are required, and the allowed values. Read it instead of guessing a channel's settings.

• list_postsFree

  Recent publishes made through us and the per-target status of each, scheduled and draft posts included (their status says so and scheduledAt says when). Newest first, 20 by default. Pass limit for more or fewer, up to 100. hasMore means the list was cut short; pass the returned nextBefore as before to keep going. To answer "what is scheduled this week", pass status scheduled with since and until and sort scheduled. Posts that already existed on the channel are not here. Use list_channel_posts for those.

• list_repliesFree

  The whole reply thread under a post, nested replies included. `truncated` tells you we stopped before the end. The count here can differ from the replies metric in get_insights, which is normal. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• media_completeFree

  Step 2 of attaching an image or video. Call it after the upload finishes. We check the file really landed before marking it ready. Only a ready media id can be passed to publish.

• media_from_urlFree

  Attach an image or video that is already on the public web. We download it, copy it into our storage and give you a media id you can pass to publish. One step, no upload needed. https only. Google Drive and Dropbox **share** links do not work: they return an HTML preview page, not the file. Use a direct file URL that ends in the file itself. If you can see the image, write altText describing it.

• media_presignFree

  Step 1 of attaching an image or video. Returns a media id and a one-time uploadUrl. PUT the file bytes to uploadUrl with the same contentType, then call media_complete. Images: image/jpeg, image/png, image/webp, image/gif, up to 20MB. Video: video/mp4, video/quicktime, video/webm, up to 8GB, 43200 seconds, 4096px wide. Aspect ratio up to 20:1. Any image pixel width is fine. If you can see the image, write altText describing it. The response has mediaExpiresAt: this media id disappears after that time if it was never published, and publish will then fail with media_expired.

• media_upload_linkwrite actionFree

  Ask the person to upload files from their own device. Returns a short-lived link. **Give the link to the person, then wait.** Poll media_upload_status with the token until it returns ready, and only then call publish with the media ids it gives you. Do not publish before the status is ready. Use this when the file is on their computer or phone; use media_from_url when the file already has a public https address.

• media_upload_statuswrite actionFree

  Has the person uploaded yet? Returns waiting, ready or expired, plus every media id uploaded through that link. Pass those ids to publish as mediaIds. Ready images also come back as image blocks so you can SEE each photo and place it in the right paragraph: previewIds[i] is the media id of the i-th image. Up to 20 images per call; pass offset to see the rest. A media with duplicateOf is the same bytes as that other id — use one of them.

• naver_layoutFree

  Naver Blog only. Shows how publish will lay the body out before anything goes out. By default (layout: template) publish reshapes the markdown into the blog's house form: #/## titles become underlined quote headings, a thin rule sits between sections, photos you did not place with media: references go one per section and the rest pair up at the end. Text never changes. Call this with the same content and mediaIds you will publish, show the person the result if they care about structure, then publish (it applies the same layout) or pass options.naver_blog.layout: "as-is" to publish exactly what you wrote. To let the person choose a form, pass forms: "all" (or a list of ids): you get every preset laid out side by side with its name, when to use it and a summary (sections, photos). Show them, let the person pick, then publish with options.naver_blog.form set to the chosen id. The server never picks for you.

• open_postwrite actionFree

  Everything about one post in a single call: the text, the whole reply thread, and its metrics. This is the right tool when someone hands you a post link. Replies or metrics can come back null if the platform refused just that part. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• publishwrite actionFree

  Post to social channels. Channels open today: threads, instagram, youtube, facebook, bluesky, telegram, naver_blog, tiktok. Get accountIds from select_channels — do not guess which channel the person meant. Naver Blog caps how many posts one ID publishes; when it does, this returns 429 naver_rate_limited with retryAfterSeconds. Do not call again before that, the draft is already in Naver. Naver Blog asks which form to publish with, every post: if options.naver_blog.form is missing this returns 400 naver_form_required with forms (id, name, when). The form is the house style to use when the person has no style of their own, so read the conversation first. If they dictated the structure or you laid it out yourself, pass options.naver_blog.layout: "as-is" and the markdown goes out unchanged. Otherwise show them the forms (naver_layout with forms: "all" lays every one out side by side) and call again with the one they pick. Do not pick for them. Naver Blog goes out through the user's browser extension. If the response carries a bridge object, read bridge.state: offline means that Chrome is closed and the post is queued (up to 7 days); logged_out means Chrome is on but not logged in to Naver. Either way relay bridge.userMessage to the person word for word, do not say it was published, and know that delete_post cancels a queued post and update_post rewrites it before it goes out. If the extension in that Chrome is too old for what you asked (update_post), this returns 422 extension_outdated with the installed and required versions; ask the person to update the extension. Text limits differ per channel: Threads 500 characters, Instagram 2200 characters, YouTube 5000 UTF-8 bytes, Facebook 63206 characters, Bluesky 300 graphemes and 3000 UTF-8 bytes, Telegram 4096 characters (1024 with media attached), Naver Blog 30000 characters, TikTok 2200 characters. Over the limit nothing goes out to any channel, so shorten it before calling. Images and video both work on the channels that take them. How several items sit in one post differs per channel: Threads groups up to 20 items in one post, Instagram groups up to 10 items in one post, YouTube takes 1 video and no images, Facebook groups up to 10 items in one post, Bluesky has no carousel and places up to 4 images in the post itself, Telegram groups up to 10 items in one post, Naver Blog has no carousel and places up to 40 images in the post itself, TikTok groups up to 35 items in one post. More than a channel takes is not refused: the first items up to its limit go out and that target's warning says what was left out. A channel that cannot mix images and video keeps the video. YouTube is different: it takes exactly one video, no images, and it needs options.youtube.title. Its description is measured in UTF-8 bytes, so Korean and Japanese cost three per character. Instagram cannot publish text alone: every post needs at least one image or video. A single video becomes a reel there. Set options.instagram.contentType to story for a story; stories show no caption. Non-JPEG images are converted for Instagram automatically. Facebook publishes to a Page, never a personal profile. options.facebook.link makes a link post (no media alongside), and a single video becomes a reel (3-90 seconds). Bluesky counts graphemes, not characters, and also caps UTF-8 bytes, so a post of 300 emoji can fail on the byte limit. Set options.bluesky.langs to the language of the text (1-3 BCP-47 codes like ["ko"]): without it the post never appears in language-scoped feeds, and Bluesky has no post editing to fix it later. Links, @mentions and #hashtags in the text are made clickable for you, and a link gets a preview card, so write the URL plainly. There are three ways to get a media id, pick by where the file is: media_from_url when it already has a public https address, media_upload_link when it is on the person's own device, media_presign plus media_complete when you can PUT the bytes yourself. Then pass the media ids here. On Naver Blog the body can also place media itself with  and @video(media:<id>), each on a line of its own. Ids you reference that way are picked up even if you leave them out of mediaIds, and media you pass but never reference goes at the end of the post. Those references only work when every target is Naver Blog: other channels would publish the markup as literal text, so we refuse instead. This publishes immediately unless you pass scheduledAt (we hold the post and send it at that time, on every channel) or draft: true (nothing goes out; the person or update_post finishes it later). scheduledAt needs a timezone offset, 10 minutes to a year out; ask the person which timezone they mean instead of guessing. A scheduled or draft post comes back with status scheduled or draft, can be changed with update_post, sent early with publish_now, and dropped with delete_post. Every channel also takes options.<channel>.content to send that channel a different text than the shared content, which is how Naver Blog Markdown and a 500-character Threads post fit in one call. On YouTube we pass privacyStatus through as you set it and report back what YouTube actually applied, so read the warning on the result instead of promising the person a visibility we did not confirm. Returns while the post is still publishing. The permalink is null at that moment. Call get_post with the returned id to see the final status and link. Pass wait: true to hold the response until it is really out — then you can tell the person it is posted instead of guessing. For a long post use threadItems instead of publish-then-reply: we keep the order and wait for each piece to land before sending the next one.

• publish_naver_draftwrite actionFree

  Naver Blog only. Publishes a draft from the blog's draft box exactly as it is in Naver: the extension loads that draft in the editor and presses publish, so edits the person made by hand in Naver are kept. Do not send content. Category, tags and openType are taken from the draft unless you pass them. Returns the same shape as publish (202 with a target that resolves to published, or the bridge object when the extension is offline; wait: true holds until it settles). If that draft was made through uplika (a target whose externalId starts with draft:), the same post record flips to published instead of a second one appearing. To publish uplika's own copy of the text rather than what is in Naver, use update_post on that post instead. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• publish_nowwrite actionFree

  Send a scheduled or draft post right now instead of waiting. Returns while it is still publishing, like publish; pass wait: true to hold for the result. A draft needs at least one target account first. Posts that already went out return post_not_editable.

• put_automationwrite actionFree

  Replace an automation's whole document. Pass the version you got from get_automation; a stale version is refused with version_conflict so a concurrent edit is not overwritten. Node types: send (mode window | private_reply | public_reply), condition, action, delay, random, goto, ai. Waiting is a send node with buttons and next: null; the tapped button's next continues. A loop must pass through such a wait. Validation problems come back with paths. Changing a live flow changes what goes out to people.

• read_conversationFree

  One conversation with its recent messages and the contact: name, whether they follow the account (Instagram only), tags, opt-out. AI drafts waiting for approval show as status draft.

• refresh_accountFree

  Re-read one connected channel's metadata. On Naver Blog this re-reads the blog's categories through the user's browser extension and waits up to a minute for it; list_accounts then shows the new list under naverBlog.categories. Call this when a category the person mentions is not in list_accounts yet. If the extension is offline you get 202 and the refresh runs when that browser comes back. Other channels return refresh_unsupported because their metadata is live on every call. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• replyFree

  Reply to a post or to a reply. Leave replyTo empty to reply to the post itself; pass a reply id from list_replies to nest a reply under that reply.Like publish, this returns before the reply is live. Call get_post with the returned id to see the final status and link. If the response carries `bridge`, follow `bridge.nextStep`: `run_wake` means run `bridge.wake` for the user's OS if you have a shell on their computer, wait about a minute, call bridge_status, then retry once, otherwise tell the user `bridge.userMessage` as it is; `ask_user_login` / `ask_user_update` mean tell the user `bridge.userMessage`. Do not retry more than twice.

• research_naver_keywordsFree

  Naver Blog only. Measures keywords (monthly searches from Search Ad, blog document count and posts per month from API HUB), judges each one (best, possible, hard, wall, hot, saturated, phantom and so on, with why), and groups them into sets for one post: a main keyword plus two to five subs with the same search intent. Every set carries prompt, a ready-to-paste Korean brief for writing the skeleton of that post. Up to 60 keywords. Measurements are cached seven days across users; new ones run against a time budget, so partial: true with unmeasured[] means the budget ran out and calling again with the same keywords finishes the rest from cache. Each call is saved as a report (reportId) unless the same keyword set was saved in the last ten minutes, which returns that report's id instead. Every set also carries draftPrompt, a brief for writing the whole post (title, subheadings, body, photo placement); pass topic and photos to fill it in. Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• retry_postwrite actionFree

  Retry the targets that failed on a publish. Targets that already went out are left alone. Naver Blog caps how many posts one ID publishes; when it does, this returns 429 naver_rate_limited with retryAfterSeconds. Do not call again before that, the draft is already in Naver. Naver Blog asks which form to publish with, every post: if options.naver_blog.form is missing this returns 400 naver_form_required with forms (id, name, when). The form is the house style to use when the person has no style of their own, so read the conversation first. If they dictated the structure or you laid it out yourself, pass options.naver_blog.layout: "as-is" and the markdown goes out unchanged. Otherwise show them the forms (naver_layout with forms: "all" lays every one out side by side) and call again with the one they pick. Do not pick for them. If nothing failed you get nothing_to_retry. Only applies to posts published through us. This replays the same payload, so read errorCode and retryable on get_post first: when retryable is false the arguments have to change and you should call publish again instead. On Naver Blog a failed target may still have left a post or a draft in the editor, so check the blog before retrying.

• search_youtube_videosFree

  Beta. Searches YouTube for a keyword and returns the top videos with views, subscribers, the views-to-subscribers ratio (above 1 means the title and topic pulled more people than the channel has), Shorts or long-form (60 seconds or less counts as a Short), duration and publish date. order is viewCount, date or relevance; period 7d, 1m, 3m, 6m or 1y; format all, shorts or long; max 1-50 (default 25). The same search is cached 24 hours and does not count; a new one spends one of the person's daily searches (quota in the response) and shared YouTube Data API units (youtube_quota_exhausted when today's are gone). Carries prompt, a Korean brief that turns the table into title and hook ideas; pass topic to fill it in. Saved to the person's search history (reportId). Returns 403 research_tool_disabled with enableUrl if the person has not turned this tool on for connectors; tell them to open that link.

• select_channelsFree

  Pick which connected channels to post to. **Call this before publish and show the result to the person.** Leave scope empty to get the candidate list and let them choose. Use scope: "all" for every active channel, or an array mixing platform names ("threads"), handles ("@vibe.trender") and account ids. Duplicates are folded, expired and not-yet-live channels are dropped into `skipped` with a reason, a sentence you can read to the person, and a link to reconnect. Pass the returned accountIds to publish unchanged. `limits` is the tightest rule across the chosen channels, so write to that. If both accountIds and candidates come back empty, nothing is connected yet. Send the person to https://uplika.com/dashboard/connections to connect a channel, then call this again. When a Naver Blog account is selected the response carries bridge with state (online, offline, logged_out), userMessage to relay to the person, and queued; if state is not online, tell the person before publishing.

• send_dmwrite actionFree

  Send a message in a conversation as the account. **Only inside the 24-hour window** after the person's last message; outside it the call is refused with window_closed and nothing can be done until they write again. On a mention thread (Threads) this posts a public reply. Pass draftId to send an AI draft that was waiting for approval. Sending pauses automations on that conversation for 30 minutes.

• update_mediawrite actionFree

  Mark or unmark an uploaded image or video as AI-made (aiGenerated). Applies to posts published or edited after this call; posts already out do not change (use update_post to rewrite a Naver Blog post). Every copy of the same file in your workspaces follows.

• update_postwrite actionFree

  Change a scheduled or draft post before it goes out: content, mediaIds, accountIds, Naver Blog caps how many posts one ID publishes; when it does, this returns 429 naver_rate_limited with retryAfterSeconds. Do not call again before that, the draft is already in Naver. Naver Blog asks which form to publish with, every post: if options.naver_blog.form is missing this returns 400 naver_form_required with forms (id, name, when). The form is the house style to use when the person has no style of their own, so read the conversation first. If they dictated the structure or you laid it out yourself, pass options.naver_blog.layout: "as-is" and the markdown goes out unchanged. Otherwise show them the forms (naver_layout with forms: "all" lays every one out side by side) and call again with the one they pick. Do not pick for them. If the Naver editor shows a "missing image" error when you open an already-published post for editing, calling update_post on that post rewrites it in place and fixes it; the URL stays the same. options, scheduledAt or draft. Fields you leave out keep their current value; options you pass replace the whole options object. scheduledAt moves the send time (same rules as publish), null turns it into a draft, and draft: true does the same. A thread (chain) can be changed too while it is scheduled or a draft: content replaces the first item and keeps the later items; threadItems replaces the whole chain (one item collapses it to a single post); passing threadItems to a single post turns it into a chain. Pass id of the first item — later items return 409 thread_piece with rootId. Posts that already went out return post_not_editable, except Naver Blog: a published Naver post is rewritten in place (same URL, same logNo) when you pass content, mediaIds or options — list the media ids again to keep the photos. A Naver draft (a target whose externalId starts with draft:, from options.naver_blog.draftOnly) is not on the blog: update_post publishes uplika's copy as a new post and, with extension 0.7.2 or later, removes the old draft from Naver's draft box. To publish the draft exactly as it is in Naver, call publish_naver_draft.

• –Code scanremote-only server, no package to scann/a
• Live reliabilityremote reachable in 1967ms20/20
• Tool poisoning54 tool descriptions checked13/15
• Auth qualityopen endpoint exposes 18 write-action tools with no auth3/15
• Maintenanceno repository listed3/15
• Maintainer identityverified namespace with website, no repo4/10

uplika: common questions

Is uplika MCP server safe?

With care: it is graded C, so read the findings first (57/100). Read the uplika safety report

How do I install uplika?

It runs remotely at api.uplika.com. Add it to Claude Code, Claude Desktop or Cursor with the snippets above, or call it through the mcp.market gateway without installing anything.

Does uplika need an API key?

Not as far as the registry entry and our scan can tell: no credentials are declared or required.

Is uplika maintained?

The latest release is v1.0.0.

Is uplika up?

95% of our last 20 checks got an answer. We check remote servers about four times a day.

What can I use instead of uplika?

Servers from other publishers that do the same job: Socialclaw MCP server, Agent-Native Chat MCP server and Godot MCP server. Compare all uplika alternatives.