Sort

AgentEarth Tool Catalog

Social Media

852Tools
12tags
14Providers

Query social media profiles, posts, notes, comments and interaction data for account research, topic tracking and content performance analysis across platforms such as X (Twitter), RedNote (Xiaohongshu), Reddit and TikTok.

  1. Find information using accounts, content URLs or keywords, and retrieve followers, likes, comments and other data through the relevant tools.
  2. Support brand monitoring, competitor account research, topic tracking and content performance analysis with data for agent research tasks.

Tool List

852 tools

ae_chanmama_video_library_search

Search Chanmama’s popular video library by keyword or topic, with date, category, audience, duration, and engagement filters. Keep the interval between create_start_time and create_end_time within 90 days per request. Split longer periods into separate requests.

ae_social_media_cross_platform_search

Search Chinese social media and public content sources by keyword across one or multiple platform categories in a single request, including Weibo, Douyin, Toutiao, Xiaohongshu, short-video sources, WeChat official accounts, news, forums, long-video sources, and news apps. Select matching fields such as title, body, author name, image/video-frame text, or audio, and specify exclusion words. You can also filter by author, content type, and sentiment. Space-separated keywords use AND matching; OR, parentheses, and quoted-phrase syntax are unverified. Dates use inclusive Beijing-time days in YYYY-MM-DD format, default to the last seven days, and support a range of at most 90 days ending no later than today. Returns unified content details, total hits, counts by platform, and pagination information, with 10 or 20 results per page and up to page 50. Complete coverage of platforms or content is not guaranteed. Successful searches with results are billed at 10.52 credits per request. Empty results are free.

ae_chanmama_author_detail

Get basic Chanmama information for a creator using their author ID, nickname, or Douyin account name.

ae_xiaohongshu_user_search

按关键词分页搜索小红书用户及创作者账号,适用于博主挖掘、账号调研及品牌相关账号查找。首次查询省略分页状态;后续页使用本工具上次返回的游标、搜索或会话状态,不能复用旧供应商的游标。

ae_bilibili_fetch_video_comments

Use this tool to retrieve comments for a specific video. Enter its video ID as `bv_id`; `pn` is an optional page number and defaults to 1. The response provides comment data for the requested page.

ae_youtube_channels_list

Find YouTube channel details by username, handle, or ID.

ae_xiaohongshu_user_get_profile

通过小红书用户ID或支持的主页链接获取用户的公开主页详细信息。

ae_social_media_search_content

Search news, Weibo, WeChat, Zhihu, Douyin, Xiaohongshu, Bilibili, and Kuaishou content for topic monitoring and discovery. Set source to ALL (the default) or one supported platform. The optional keyword supports && for AND, || for OR, - for exclusion, parentheses for grouping, and English double quotes for exact matching without tokenization. On the first request, provide both start and end in yyyy-MM-dd HH:mm:ss format. To retrieve another page, pass nextCursor from an earlier response when available. The response contains search output in data together with code, message, and recordTime; the detailed structure of individual results is not specified. Results should not be treated as guaranteed complete or real-time coverage, and each call is a search rather than a persistent monitoring subscription.

ae_douyin_fetch_video_statistics

Retrieve statistics for one or two Douyin videos. Input: aweme_ids is required and must be a string containing a single video ID or two video IDs separated by a comma. A maximum of two IDs is supported per request. Returns: Video statistics, including documented metrics for likes (digg_count), downloads (download_count), plays (play_count), and shares (share_count). The statistics are returned in the response's data field; its detailed structure is not specified. Example: aweme_ids = "7448118827402972455,7126745726494821640".

ae_wechat_get_account_articles

### Purpose Retrieve one page of historical posts from a WeChat official account. Articles are selected by default; video, audio, and image-text tabs are also supported. Each call returns a single page and does not automatically retrieve the account's full history. ### Inputs Provide a required `request_body` object containing: - `username` (required): the account identifier, such as `gh_363b924965e9`. Supported forms include `gh_…`, `gh_…@app` for mini-program-linked accounts, and custom WeChat IDs such as `nikejdi`. - `offset` (optional): a base64 pagination cursor. Omit it or use an empty string for the first page. For the next page, pass the previous response's `data.next_offset` unchanged. - `page_size` (optional): integer from 1 to 100, default 20. WeChat currently ignores this value; the actual number of returned items is determined by the account. Use the cursor to paginate. - `item_show_type` (optional): omitted or `0` for articles, `5` for videos, `7` for audio, or `8` for image-text posts. - `raw` (optional): defaults to `true` for raw nested batch entries. Set `false` for simplified article records, recommended for list display. ### Results and pagination The response's `data` contains `biz_username` (account identifier), `articles` (this page's records), `count` (number returned on this page), `is_end` (0/1 end-of-list flag), and `next_offset` (cursor for the next request; null on the last page). Stop paging when `is_end` is `1`. With `raw=false`, records in `data.articles[]` use snake_case fields including `app_msg_id`, `title`, `digest`, `url`, `cover`, `covers`, `create_time`, and `update_time`, along with position, type, image-count, and paid-content indicators. With `raw=true`, records are full raw batch entries containing `appMsg` and `baseInfo`; article content is under `data.articles[N].appMsg.detailInfo`, and one batch can contain multiple articles. Both formats use the same top-level pagination fields. Responses can be slow; a 30-second client request timeout is recommended. Timeout is a client setting, not an input parameter.

ae_douyin_tiktok_parse_single_video

Parse a single Douyin or TikTok video from a video link, share link, or share text. Inputs: - url (required string): The video link, share link, or share text for one video. - minimal (optional boolean, default false): Whether to return minimal data. - base64_url (optional boolean, default false): Whether the URL is submitted using Base64 encoding. Returns: A direct response with video data in data. The data value may be null, and its detailed fields are not specified. No task polling is required.

ae_youtube_commentthreads_list

List top-level comment threads for a specific YouTube video or channel.

ae_xiaohongshu_get_keyword_stat

## Xiaohongshu keyword overview Look up a keyword's site-wide popularity and the aggregate profile of cooperating bloggers to support keyword comparison and content planning. ### Input Provide `request_body.search_word` as a string of 1–100 characters, for example: ```json {"request_body":{"search_word":"面膜"}} ``` The request body accepts only `search_word`. ### Results The business payload is located at `response.data.data` when available: - `searchWordData`: `searchScore` is the search index, `noteNum` is the related note count, `userNum` is the number of search users, and `readNum` is the read count. Each metric has a corresponding `*30bm` field describing its 30-day percentage change. - `wordKolProfile`: aggregate content-category and persona shares for cooperating bloggers. This is not a list of individual bloggers or a demographic profile of people searching the keyword. A null payload indicates no result; it should not be treated as zero for every metric. The tool returns an overview, not daily trends or related-keyword lists. Search index and note count can help compare interest with content supply, but do not by themselves establish a keyword's competitiveness.

ae_douyin_index_fetch_item_query

Search the Douyin Index video library by keyword and optionally narrow results by category, publication time, video quality label, and duration. Inputs: - query: Required search keyword (string), for example "food". - category_id: Optional category ID (string); default "0" means all categories. Examples: "628" = Food, "629" = Travel, "615" = Technology. - date_type: Optional integer publication filter; 0 = any time (default), 3 = last 3 days, 7 = last 7 days, 30 = last month. - label_type: Optional integer quality label; 0 = any (default), 1 = low-follower hit, 2 = high completion rate, 3 = high follower growth, 4 = high like rate. - duration_type: Optional integer duration filter; 0 = any (default), 1 = 0–15 seconds, 6 = 15–60 seconds, 7 = 60–120 seconds, 8 = 120–180 seconds, 9 = over 180 seconds. Current filter choices can be retrieved through the filter-options lookup when available; category examples above are not a complete list. Returns a response with search results in data. Documented video information includes ID, title, cover, duration, publication time, play count, likes, comments, shares, and author ID, nickname, and follower count. The result data has no fixed detailed schema, so available fields depend on the returned records. Results cover the Douyin Index video library, not necessarily all Douyin videos. Example: {"query":"food","category_id":"628","date_type":7,"duration_type":6,"label_type":2} searches the Food category for matching videos published in the last 7 days, lasting 15–60 seconds, with the high-completion-rate label.

ae_youtube_get_channel_videos

## Get YouTube channel videos Retrieve one page of a channel's video listing, with optional language and region preferences. Use the returned pagination token to request additional pages. ### Inputs - **channel_id** (required string): The actual YouTube channel ID, for example `UCJHBJ7F-nAIlMGolm0Hu4vg`. Extract it from a `/channel/ID` URL. If you only have an `@handle`, obtain its channel ID first; do not pass the handle or full URL as this value. - **language_code** (optional string; default `zh-CN`): Language preference, such as `en-US`, `ja-JP`, or `ko-KR`. - **country_code** (optional string; default `US`): Region code, such as `US`, `JP`, or `GB`. - **continuation_token** (optional string): Copy the token from the previous response to retrieve the next page. Omit it for the first page. - **need_format** (optional boolean; default `true`): Return cleaned, simplified listing data when `true`, or raw data when `false`. ### Results The response's `data` contains the listing. With `need_format=true`: - **videos**: Video entries containing IDs, titles, thumbnail links, duration and accessibility text, view-count text, published-time text, live status, verification information, video-page URLs, and playback initialization URLs. - **channel**: First-page channel metadata, including ID, name, bio (`description`), handle URL, avatar, homepage and RSS links, keywords, family-safety status, and verification status. - **continuation_token**: The token for requesting another page when available. A single call does not retrieve all channel videos. ### Important limitations Save the first page's `channel` metadata: later pages omit it. On later pages, `videos[].is_verified` is always `false` and is not a reliable verification indicator; use the first page's channel-level value instead. In the documented lockup listing format, `videos[].description` is always `""`; use `channel.description` for the channel bio and separate video-detail retrieval for individual video descriptions. In that format, `short_view_count` duplicates `view_count` for compatibility. View counts and published times are display text rather than normalized numeric counts or timestamps.

ae_wechat_get_video_detail

## WeChat Channels video details Retrieve one video's details, including its ID, creator, title, engagement counts, publication information, media URLs and decryption key. ### Inputs Provide a `request_body` object containing at least one nonempty video locator: - `object_id`: Video objectId as a numeric string; highest priority. - `export_id`: Search result exportId beginning with `export/`; used when `object_id` is empty. It expires, so use it promptly. - `share_url`: Channels share short link such as `https://weixin.qq.com/sph/AH3sCoIPhH`; used only when both IDs above are empty. Priority is `object_id > export_id > share_url`. Optional `object_nonce_id` is a numeric-string nonce, available as `feedNonceId` in search results, that can improve matching alongside an ID. Optional `raw` defaults to `true`; set it to `false` for a simplified result, recommended for media download workflows. ### Returned data - With `raw=false`, `data` is one video object with `id`, `username`, `nickname`, `title`, engagement counts, `create_time`, `object_type`, `location` and `media`. The media object contains `url`, `url_token`, the combined `full_url`, and `decode_key`. - With `raw=true`, the video is at `data.objects[0]`. Media is at `data.objects[0].objectDesc.media[0]`, with `url`, `urlToken` and `decodeKey`. The nested status is at `data.baseResponse.ret`. This tool returns information, not a downloaded or decrypted file. Use `full_url`, or concatenate the media URL and token, to obtain the complete media URL. Downloaded video may still require decryption; the encrypted file and key must come from the same response because every request returns a new link and key. Preserve large IDs as strings with lossless JSON parsing rather than JavaScript `Number`. Responses may be slow; set the client request timeout to 30 seconds.