Skip to main content
Glama

Server Details

Make social content from your agent: hooks, AI UGC videos, competitor ads, posts to 30+ networks.

Ownership verified
Status
Healthy
Uptime
100.0% over the last 24 hours
Last Tested
Transport
Streamable HTTP ยท MCP 2025-11-25
URL
ChatGPT
โ€“
Claude
โ€“

TDQS

B3.1/5.0

Scored across 36 tools

Disambiguation2/5

Several tools overlap heavily: generateImageTool/generateVideoTool duplicate studio_generate for media creation, while videoFunctionTool/generateVideoOptions and videoStatusTool parallel studio_formats/studio_status. Upload helpers (uploadWidgetTool, uploadWidgetStatusTool, uploadWidgetTicketTool, uploadFromUrlTool, mediaListTool) also have fuzzy boundaries for an agent choosing the right path.

Naming Consistency2/5

The set mixes snake_case (brand_profile, hooks_generate, sales_overview, studio_generate) with camelCase and a 'Tool' suffix (deletePostTool, integrationSchedulePostTool, uploadFromUrlTool, videoStatusTool). Some names are bare nouns (benchmarks, groupList) while others are verb_noun, so the convention is not predictable.

Tool Count2/5

36 tools is heavy for this surface, especially because it includes internal widget helpers (uploadWidgetTicketTool), legacy video-generation helpers, and parallel Studio tools. A tighter set could cover the same workflows with fewer, clearer tools.

Completeness4/5

The surface covers the core lifecycle well: brand reading/profile, hooks, strategy, Studio generation, media upload/list, post scheduling/listing/settings/deletion, sales tracking, radar, benchmarks and credits. Minor gaps remain, such as updating post content after creation and managing/deleting media library items.

Available Tools

36 tools
ask_hookbirdBInspect

Ask agent 'hookbird' a question. Agent description: Hookbird agent: reads the brand, writes and scores hooks, makes content in Studio and schedules it to social channels

Inputs: message (required).

benchmarksA
Read-onlyIdempotent
Inspect

Compare this account's last 30 days (engagement rate, posting frequency and more, from connected channels) with the niche median and top 10%, per platform. Free. Every metric carries its source report and year, and a plain "why" sentence. Top 10% and percentiles marked estimated are estimates: say so. "you" is null when that channel is not connected. Use it when the user asks "how am I doing?" or "what's good for my niche?".

Inputs: niche, platform. Returns: error, niche, method, platforms, +1 more

brand_profileA
Read-onlyIdempotent
Inspect

Get the brand Hookbird knows for this account: name, website, niche, voice, products, positioning, competitors and audiences (each audience has an id, what they want and what they are stuck on). "kind" says whether it is a brand or a creator (one person; their audiences are viewers). "socials" are the Instagram, TikTok and Facebook accounts it read, and "whatWorks" their posts that beat the account's usual (lift = times its median), with why and the hook as a reusable pattern. Call this first in any content task: hooks, angles, plans and Studio content are all written from it. Pass an audience id to hooks_generate, hooks_score and strategy_plan to aim at one audience. When "found" is false the brand was never read: ask for the website or the Instagram or TikTok account and call brand_read.

Inputs: no named parameters. Returns: error, found, profile

brand_readA
Destructive
Inspect

Read a brand from its website and its Instagram, TikTok or Facebook accounts, and save it as the brand profile: name, logo, colours, voice, products, audiences, competitors and market white space. Free, takes 1 to 3 minutes. Start it with "url", "socials" or both: a creator or a shop with no website is read from its accounts alone. It answers with a jobId: call this tool again with only "jobId" to follow it (each call waits up to 20 seconds for progress). When status is "done", call brand_profile. Reading again replaces the current profile, so only do it when the user asks or when brand_profile has nothing.

Inputs: url, jobId, socials. Returns: step, error, jobId, label, +3 more

credits_balanceA
Read-onlyIdempotent
Inspect

How many Hookbird credits this account has left: plan credits (refilled each billing period, spent first), extra credits from packs (never expire), the total, the monthly allowance and when the period ends. Studio generations spend credits (prices in studio_formats); hooks, brand, plans, Radar and benchmarks are free. Check this before a large generation or when the user asks about credits.

Inputs: no named parameters. Returns: plan, error, extra, total, +2 more

deletePostToolA
DestructiveIdempotent
Inspect

Use this when the user wants to delete an existing post (scheduled, draft, published or errored). Find the post first (list your posts) and pass its "id" here. It deletes the post on that one channel together with its comments - a post created for several channels has a separate id on each channel. A repeating post is listed once per repetition with the same id - deleting it deletes every repetition. A deleted scheduled post is not published. Deleting a published post only removes it from Hookbird - it stays live on the social network. The deletion cannot be undone. If the post cannot be deleted, the result contains output.errors explaining why.

Inputs: id (required). Returns: output

generateImageToolCInspect

Generate image to use in a post, in case the user specified a platform that requires attachment and attachment was not provided, ask if they want to generate a picture of a video.

Inputs: prompt (required). Returns: id, path, error

generateVideoOptionsC
Read-onlyIdempotent
Inspect

All the options to generate videos, some tools might require another call to generateVideoFunction

Inputs: reasoning. Returns: video

generateVideoToolBInspect

Generate video to use in a post, in case the user specified a platform that requires attachment and attachment was not provided, ask if they want to generate a picture of a video. In many cases 'videoFunctionTool' will need to be called first, to get things like voice id Generating takes a few minutes, so this only starts the generation and returns a jobId: poll 'videoStatusTool' with it until the status is "completed" to get the video url. Here are the type of video that can be generated:

Inputs: output (required), identifier (required), customParams (required). Returns: error, jobId

groupListA
Read-onlyIdempotent
Inspect

This tool lists the available groups (customers). Use a group id with the integrationList tool to filter the integrations belonging to that group

Inputs: no named parameters. Returns: output

hooks_generateAInspect

Write hooks (the first line or first 2 seconds of a post) for this brand, each scored 0-100 with its pattern, an estimated percentile and why it works. Best first. Free (no credits). Written from the brand profile when there is one; without one (brand_profile finds nothing) give a topic and they are written for it. Aim it with a topic, an audience id from brand_profile and the platform. Pick the best hook and pass it as "hook" to studio_generate, or use it as the first line of a post.

Inputs: count, topic, platform, audienceId. Returns: error, hooks, audience

hooks_scoreA
Idempotent
Inspect

Score one hook (the opening line of a post or video) from 0 to 100. Free (no credits). Returns the pattern it uses, its strengths, concrete fixes, the score breakdown and an estimated percentile with the basis it was estimated on (say "estimated" when you quote it). "rewrite" is a stronger version that was scored too ("rewriteScore") and beats the hook without being longer; when there is none, no version we wrote scored higher: say the hook already holds up and give the fixes, never invent a rewrite. Use it when the user asks "is this hook good?", or to check a hook before generating content with it. Works without a brand profile, but is sharper with one.

Inputs: text (required), save, platform, audienceId. Returns: id, text, error, fixes, +7 more

integrationListA
Read-onlyIdempotent
Inspect

This tool list available integrations to schedule posts to. Optionally pass a group id (from the groupList tool) to only list integrations belonging to that group

Inputs: group. Returns: output

integrationSchedulePostToolA
Destructive
Inspect

Use this when the user wants to create a draft, scheduled, or immediate social media post on their connected channels, based on the integrationSchema tool. Examples of the input shape:

A single LinkedIn post with one comment

  • socialPost array length will be one

  • postsAndComments array length will be two (one for the post, one for the comment)

20 Facebook posts each on individual days without comments

  • socialPost array length will be 20

  • postsAndComments array length will be one

Do not use this to update or delete existing posts. On success, each item in output contains a previewUrl the user can open to see the post. If validation fails, the result contains output.errors describing what to fix; the call can be retried with corrected parameters.

Inputs: socialPost (required). Returns: output

integrationSchemaA
Read-onlyIdempotent
Inspect

Returns the post settings a channel's network needs before a post can be scheduled to it (for example a board, a subreddit or a title) and the lookup helpers it offers. Some settings need an id from the network (a Pinterest board, a subreddit, a page): the response lists the helpers under 'tools'; run one with triggerTool to get the id. 'tools' can also include a mention lookup, to tag people or pages in the post content.

Inputs: platform (required), isPremium (required). Returns: output

mediaListToolA
Read-onlyIdempotent
Inspect

List the media (images and videos) already uploaded to the media library, newest first, 18 per page, optionally searched by original filename. Use the returned path as an attachment URL in integrationSchedulePostTool instead of uploading the file again.

Inputs: page, search. Returns: pages, output

postSettingsToolA
Destructive
Inspect

Update the provider settings of an existing post (scheduled or draft) that was NOT published yet. Only the settings change - the content and the publish date stay exactly as they are. Find the post first (list your posts) and pass its "id" here. The settings are merged into the existing ones, so only pass the keys you want to change; anything you don't pass stays as it is. This relies on the integrationSchema tool [input:settings] to know which keys exist for the platform. If validation fails, the result contains output.errors describing what to fix; the call can be retried with corrected parameters.

Inputs: id (required), settings (required). Returns: output

postsListToolA
Read-onlyIdempotent
Inspect

List the organization's posts scheduled to be published between two dates (the same data as the "List Posts" API endpoint). Returns every post in the window whatever its state (scheduled, draft, published, errored). "startDate" and "endDate" are required (UTC) - to list all upcoming posts, pass a wide window (for example from now to a year ahead). Each item has an "id", its publish date, state, error (for errored posts), content, media (the attached images and videos: id, path, and thumbnail / alt / type when set), channel and current provider settings.

Inputs: endDate (required), startDate (required), customer. Returns: output

radar_adsA
Read-onlyIdempotent
Inspect

Find the ads a competitor (or any search term) is running, from the ad libraries Hookbird is connected to (cached 24 hours). Free. Each ad has its hook text, platform, how many days it has been running (long-running ads are the ones that make money) and a short why. "providers" says which sources are on and, when one is off, why. When "ads" is empty and no provider is enabled, tell the user ad sources are not set up, do not invent ads. When "note" is set, no source covers the asked market: say so and that the ads shown ran elsewhere (each ad's "country"). Competitor names are in brand_profile. Use winning hooks as inspiration for hooks_generate (topic: "remix: ..."). Each ad says whether it is the competitor's own ("own": its advertiser carries the competitor's name and, when brand_profile lists the competitor's site, its ads land there) or only mentions the search word (another advertiser), and whether its text is a template ("templated": a dynamic ad whose words are filled per viewer, like "{{product.brand}}", so they are not public). Only cite, compare or remix the competitor's own ads with real text; say plainly when there are none.

Inputs: q, limit, country, minDays, +2 more. Returns: ads, note, error, query, +3 more

sales_codeAInspect

Make a discount code for a post or video in the owner's connected store, with the store's default discount (set on the Sales page, 10% off unless changed), or register a code they already have (existing true: nothing is made in the store). Each order with the code counts for that post on the Sales page: on TikTok and Instagram, where captions have no clickable links, a code said in the video or written in the caption is the surest way to count sales. No credits. It creates a real discount in the merchant's store: use it only when the owner asks for a code, and say what it gives. Clients that can ask their user ask before it runs; otherwise get a clear yes first. Pass the post's "id" (list your posts) when the post exists. For a video not posted yet, leave postId out: the code is tied to the first post saved with it in its text. Without a code, a random one is made (prefix puts letters before it, e.g. the brand). Only workspace admins make codes in the app. A store that cannot make codes from Hookbird (a website with the pixel) takes existing codes only.

Inputs: code, postId, prefix, storeId, +1 more. Returns: code, error, store, postId, +1 more

sales_overviewA
Read-onlyIdempotent
Inspect

What the brand's posts sold in its own store over the last 7, 30 or 90 days: the numbers of Hookbird's Sales page. Free. It has the revenue and orders that came from posts and their share of all the store's sales, the clicks on tracked links and the rate that became orders, the posts that sold best (hook, hook pattern, format, presenter, networks; the same post on several networks counts once), sales per network, format, hook pattern and presenter, how orders were matched to posts (matchedBy: "code" a discount code made for a video, "link" the last tracked link visited before the order, "first_link" an earlier one, "pixel" the website pixel), the latest orders and the connected stores with their status. Use it when the user asks what sold, which posts or videos make money, or which hooks, formats, presenters or networks bring orders, and before suggesting content that should sell. sales_post has one post in detail. The numbers come only from the store the owner connected (Shopify, WooCommerce, Stripe or a website with the Hookbird pixel), in "currency" (other currencies are listed apart). "connected" false means no store is connected yet: say so and that they connect one on the Sales page (salesPage); a store type can show there as not available yet. A store with status "error" says why in lastError. No sales is a real answer: never make up, estimate or promise sales.

Inputs: days. Returns: days, error, stores, totals, +12 more

sales_postA
Read-onlyIdempotent
Inspect

What one post sold in the owner's store since it was made, the post's Sales panel in Hookbird. Free. The same post on several networks counts as one: each network has its clicks, orders and revenue and its tracked links (in the post, or "bio" for a bio or story link), then the post's discount codes with their orders, and its latest orders (matchedBy: "code", "link", "first_link" or "pixel"). Pass the "id" of a post (list your posts, or a postId from sales_overview). Use it when the user asks how a given post or video sells, or for its links and codes. The numbers come only from the store the owner connected; "connected" false means no store is connected yet (they connect one on the Sales page). Never make up or promise sales.

Inputs: postId (required). Returns: codes, error, title, postId, +7 more

strategy_planAInspect

Create or read a content plan: a calendar of posts for one audience, each with a date, channel, Studio format, angle, hook, caption and brief. Free (no credits); making each post later in Studio costs its format's credits. action "create": builds a plan over 1-8 weeks with "cadence" posts a week. "channels" takes platform names ("tiktok", "instagram") and/or channel ids from integrationList; channel ids also get DRAFT posts in the calendar (nothing is published). Angles are written first when the audience has none. Takes about a minute. Confirm audience, channels, cadence and weeks with the user first. action "get": with planId returns that plan and its items, without planId lists the plans. To make an item, pass its hook and formatId to studio_generate.

Inputs: action (required), goal, goals, weeks, +5 more. Returns: plan, error, plans

studio_formatsA
Read-onlyIdempotent
Inspect

List everything Studio can generate: wall-of-text posts, TikTok slideshows, AI UGC talking heads (selfie, street interview, car yapper, podcast), explainers (a presenter teaching one idea, cut with animated graphics on the spoken words), animated stories, hook + demo videos, voiceovers, images, video clips and consistent AI influencers (avatars). Every format has a credit price per duration (formats with "qualities" also have a Premium price per duration, made with a more realistic model), the inputs it needs (a demo video, a product image or an avatar from the media library) and what it outputs. Also returns the hook formulas that can be passed as "hook" to studio_generate, and the organization's saved avatars. Call this before studio_generate, tell the user the price in credits, and get their confirmation before generating.

Inputs: reasoning. Returns: hooks, avatars, formats

studio_generateAInspect

Generate a piece of content with a Studio format (see studio_formats for the formats, their credit prices and the inputs they need). It spends the credits shown by studio_formats for the chosen duration (and quality, where the format has a Standard and a Premium price): always tell the user the price and get their confirmation first. The script, on-screen text and visuals are written from the brand profile and what you pass (hook, angle, prompt, or the user's own script). Generation runs in the background (seconds for images and text posts, a few minutes for videos and talking heads): this returns a generationId, then call studio_status with it. Results are saved to the media library and can be attached to posts. If the generation fails, the credits are given back. To use one of the user's saved recipes (studio_recipes), pass its recipeId: its format and settings are filled in, what you pass yourself wins. Pass formatId or recipeId. To make a finished generation again ("the same in Premium", "again at 30 s"), pass its id as fromGenerationId with only what changes: its format, words and settings are reused. Made for a post of the user's content plan (the chat names its planItemId): pass planItemId, the result becomes that post's content when it has none yet and is in the post's format (linkedToPlan says whether it did).

Inputs: ad, hook, angle, music, +19 more. Returns: error, credits, generationId, linkedToPlan

studio_recipesA
Read-onlyIdempotent
Inspect

List the Studio recipes the organization saved: a named format plus its settings (avatar, voice, text style, captions, aspect, duration, slides, quality, hook rule, chat app, call to action, music, script structure). To make content "the usual way" or with a named recipe, pass its id as recipeId to studio_generate: the recipe's format and settings are used, and anything you pass yourself wins. A recipe never holds the topic, hook words or media: still pass prompt / hook / mediaIds as needed. Free, no credits.

Inputs: no named parameters. Returns: recipes

studio_statusA
Read-onlyIdempotent
Inspect

Check a generation started with studio_generate, using its generationId. While it runs this call waits up to 25 seconds for a change before answering, so there is no need to wait between calls: when the status is "queued" or "running" call it again, and after a few calls tell the user it is still running. When the status is "done" the outputs are media library files (images, slides, videos) with hosted urls that can be attached to posts, plus a suggested post caption. When the status is "failed" nothing was made, the error says why, and the credits were given back.

Inputs: generationId (required). Returns: step, error, status, caption, +3 more

triggerToolAInspect

Looks up data on a connected channel's network that a post setting needs, like a Pinterest board id, a subreddit or a page to mention. It only runs one of the lookup helpers that integrationSchema lists for that channel (methodName); any other name is refused. Some helpers take a value from the user, like a word to search for [input:callable-tools]. It reads from the network and never posts or deletes anything. If the channel's sign-in has expired, the stored token is refreshed; if that fails, the channel is flagged for reconnect.

Inputs: dataSchema (required), methodName (required), integrationId (required). Returns: output

uploadFromUrlToolAInspect

Upload a remote image or video into the media library from a public http(s) URL. Local or private files cannot be fetched by this tool: upload them with the public API (POST /public/v1/upload with an API key, https://hookbird.ai/docs/public-api/introduction). Use this before scheduling a post when the user provides an external media URL (not already hosted on our domain), so the attachment passes the upload-domain validation. Returns the hosted media { id, path } to use as an attachment (or as the src of an in the content, when the channel rules say pictures go inside the content), or { error } on failure.

Inputs: url (required). Returns: id, path, error

uploadWidgetStatusToolA
Read-onlyIdempotent
Inspect

List the media the user uploaded with the upload widget, using the sessionId returned by uploadWidgetTool. It lists only the files uploaded in that widget session, not the whole media library. An empty list means the user did not upload anything yet. A media with the status "processing" is still being prepared: wait about 10 seconds and call again. A media with the status "ready" can be used as a post attachment with its { id, path }.

Inputs: sessionId (required). Returns: error, media

uploadWidgetTicketToolAInspect

Used by the upload widget to get a short-lived upload ticket for the sessionId returned by uploadWidgetTool.

Inputs: sessionId (required). Returns: error, ticket

uploadWidgetToolAInspect

Show the user an upload widget to add images or videos from their own device to the media library. Use this when the user wants to attach a local file to a post. When the media is already available on a public URL, use uploadFromUrlTool instead. The widget is only displayed by apps that support MCP Apps (interactive UI); in any other app no widget appears, and local files go through the public API (POST /public/v1/upload with an API key, https://hookbird.ai/docs/public-api/introduction). uploadFromUrlTool is only for public URLs. Returns a sessionId: the files the user uploads are reported in the conversation, and can also be read with uploadWidgetStatusTool.

Inputs: no named parameters. Returns: error, sessionId

videoFunctionToolD
Read-onlyIdempotent
Inspect

Sometimes when we want to generate videos we might need to get some additional information like voice_id, etc

Inputs: identifier (required), functionName (required). Returns: object

videoStatusToolA
Read-onlyIdempotent
Inspect

Check the status of a video generation started with 'generateVideoTool', using the jobId it returned. Generating a video takes a few minutes: while the status is "pending", wait about 30 seconds and call again. When the status is "completed" the result contains the hosted video url, which can be used as a post attachment. When the status is "failed" the result contains the error message.

Inputs: jobId (required). Returns: id, url, error, status

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelogโ€ฆ

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      D
      maintenance
      AI-powered social media posting across 14 platforms. Post to Twitter, Instagram, TikTok, Facebook, LinkedIn, YouTube and more with one command. AI adapts content per platform, schedules posts, and generates 30-day content calendars.
      6
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Social media scheduling and publishing for AI agents. 17 validation-first tools to post to X, LinkedIn, Instagram, TikTok, YouTube, Reddit, Discord, Telegram, and more through one connected workspace.
      164 npm
      110
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      Enables AI agents to publish and schedule social posts across Instagram, Facebook, TikTok, YouTube and more, analyze performance and audience metrics, and manage comments and messages with approval safeguards. It also supports short links with click tracking.
      6
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      One prompt, a native post per network, published or scheduled by your agent: TikTok, Instagram, Facebook, YouTube, LinkedIn, Bluesky, Mastodon, Telegram, Discord, Slack, Nostr, X, WordPress, Dev.to, Hashnode. 22 tools incl. plan_week, bulk_schedule, analytics and TikTok video quality check. Priced per brand, not per channel; API + MCP on the free plan. Hosted: postwire.io/api/mcp
      22
      87 npm
      MIT
    Try in Browser

    Glama MCP Gateway

    Add one secure layer between your agents and this server.

    Resources