ae_zhihu_search_discover
Retrieves Zhihu search discovery recommendations. No input is required. The result contains the available discovery data and may include response metadata such as status, request details, generation time, documentation, or cache information; the specific data fields depend on the response.
ae_zhihu_user_search
Search for users by keyword. `keyword` is required; `offset` optionally sets the starting position and defaults to `0`, while `limit` optionally sets the number of users per page and defaults to `25`. The synchronous response includes matching results in `data` along with response metadata.
ae_kuaishou_app_live_search
# [中文]
### 用途
按关键词查找当前正在直播的直播间,并支持游标分页。
### 输入
- `keyword`:必填,搜索关键词。
- `pcursor`:可选,分页游标。首次搜索时留空或省略;继续翻页时传入上一页响应中的 `pcursor`。
### 返回
返回与关键词匹配的直播间搜索结果;如需继续获取结果,可使用响应中的分页游标。
# [English]
### Purpose
Find currently live rooms matching a keyword, with cursor-based pagination.
### Inputs
- `keyword`: Required search keyword.
- `pcursor`: Optional pagination cursor. Leave it empty or omit it for the first search; for the next page, provide the `pcursor` from the previous response.
### Returns
Returns live-room search results matching the keyword. Use the pagination cursor in the response to request more results.
ae_wechat_get_channel_info
## WeChat Channels account information
Retrieve an account’s “More Info” sections: basic information such as IP location, profile location and public Channel ID (sph ID), and verification information such as service unit, entity type, internet news service license number, service category and verification time. Interpret the fields actually returned for the account.
### Inputs
- `request_body` (required): the request object.
- `request_body.username` (required): the account’s finder username, matching `^v2_[0-9a-fA-F]+@finder$` and 10–256 characters long. A display name or `sph…` Channel ID cannot be used directly. The finder username can be obtained from the `username` field in video detail or account video responses; an sph ID must first be converted to a finder username.
- `request_body.raw` (optional; default `true`): `true` returns the original response; `false` returns a simplified parsed structure.
### Results
- With `raw=false`, `$.data.finder_username` identifies the account, `$.data.channel_id` is its public sph ID, and `$.data.sections` preserves the sections. `$.data.info` is a flattened title-to-content dictionary, with keys such as `视频号ID`, `IP归属地`, `服务单位`, `主体类型`, `互联网新闻信息服务许可证编号`, `服务类别` and `认证时间`. Rich-text image markers are stripped during parsing.
- With `raw=true`, `$.data.baseResponse.ret` is the result status (`0` means success). Read section titles at `$.data.sections[N].title` and item labels and values at `$.data.sections[N].items[M].title` and `.content`; items may also contain `.type` and `.jumpInfo`. This mode does not include the flattened `info` dictionary or a separate `channel_id`; extract them from section items if needed.
Responses can be slow. A 30-second client request timeout is recommended; timeout is a client setting, not a tool input.
ae_wechat_search_channel_videos
Search a specific creator's published WeChat Channels videos by keyword. The search is limited to that account, and each call returns one page.
Inputs inside the required request_body:
- username (required): the creator's finder username, matching ^v2_[0-9a-fA-F]+@finder$, 10–256 characters. A display name is not a substitute.
- keyword (required): the search text, 1–100 characters after trimming whitespace.
- raw (optional, default true): true returns the raw grouped search response; false returns a simplified, flat hit list.
- cursor (optional): omit or leave empty for the first page. For subsequent pages, copy the previous response's next_cursor exactly. It is an opaque string of at most 4096 characters; null is also accepted. Do not parse, concatenate, or construct it.
Results:
- With raw=false, data contains username and keyword echoes, count (hits on this page), items, continue_flag, and next_cursor. Each item includes type, doc_id, and title with search highlighting removed; desc, thumb_url, username, and nickname may be null.
- With raw=true, data.query echoes the keyword, data.offset is the next-page offset, and hits are nested under data.data[N].subBoxes[M].items or data.data[N].items. The continuation field is data.continueFlag, rather than data.continue_flag.
- Both formats provide data.next_cursor. Request another page only when the continuation flag is 1 and next_cursor is non-empty, passing it verbatim as cursor. next_cursor is null when there is no next page.
Keep 64-bit IDs such as doc_id lossless by handling them as strings with an appropriate JSON parser; do not pass them through JavaScript Number. This tool returns search hits, not media download URLs or decoding keys; those require a separate video-detail lookup. Responses may be slow, so set a 30-second client timeout where configurable; timeout is not an input parameter.
ae_reddit_app_fetch_popular_feed
Fetch Reddit's site-wide popular feed for browsing popular or trending posts.
All inputs are optional:
- sort: BEST (default), HOT, NEW, TOP, CONTROVERSIAL, or RISING.
- time: ALL (default), HOUR, DAY, WEEK, MONTH, or YEAR.
- after: Pagination cursor, default empty string for the first request. Use the returned after cursor to request the next page when available; maximum 4096 characters.
- filter_posts: Post IDs to exclude to help avoid duplicates, default []. Accepts up to 100 IDs in the form t3_ followed by lowercase letters or digits, with a total length of 4–20 characters per ID.
- language: Preferred response language as a language tag such as en-US (default). Length: 2–16 characters; starts with 2–3 letters and may include hyphen-separated segments of 2–8 letters or digits.
- need_format: Whether to request cleaned/formatted data; default false.
Returns a JSON response whose data payload is documented to contain popular posts, details such as titles, content, upvotes and comment counts, and pagination information. The payload's detailed structure is not fixed by the response schema and it may be null; available post fields and a next-page cursor can vary.
Example: {"sort":"HOT","time":"DAY","filter_posts":[],"after":"","language":"en-US","need_format":false}
ae_reddit_app_fetch_post_details_batch
Fetch details for up to five Reddit posts by ID, optionally including context for a specific comment.
Inputs:
- post_ids (required): A comma-separated string of 1–5 post IDs, for example "t3_1ojnh50,t3_1ok432f". Each ID must start with t3_ followed by 1–17 lowercase letters or digits. The full string is limited to 104 characters; do not supply URLs or an array.
- include_comment_id: Set to true to request context for the comment specified by comment_id. Default: false.
- comment_id: When requesting comment context, supply a t1_-prefixed ID followed by 1–17 lowercase letters or digits. Default: empty string.
- need_format: Whether to clean and format the returned data. Default: false.
- language: Preferred response language as an IETF language tag, such as en-US; 2–16 characters. Default: en-US.
Returns: A response with post details in data. Available details may include title, content, author, vote and comment statistics, subreddit information, awards, media, and recommendation reasons. The data structure is not fixed by the documented schema and may be null; not every detail is guaranteed for every post.
Limits: More than five posts in one request returns an error. Preserve the t3_ post and t1_ comment prefixes. This is a lookup by known IDs, not a post search or a guarantee of a complete comment thread.
ae_reddit_app_fetch_subreddit_info
Retrieve detailed information about a Reddit subreddit, such as its description, subscriber count, creation time, and rules when available.
Inputs (all optional):
- subreddit_name: Bare subreddit name without r/, for example "pics" or "AskReddit". Use 2–21 ASCII letters, digits, or underscores. Default: "pics".
- language: Preferred response language as an IETF language tag, such as "en-US". Must be 2–16 characters and match ^[A-Za-z]{2,3}(?:-[A-Za-z0-9]{2,8})*$. Default: "en-US".
- need_format: Whether to request data cleaning and formatting. Default: false.
Returns: A JSON response whose data contains the subreddit metadata. The exact fields are not fixed by the documented schema, may vary, and data may be null. This tool retrieves community information rather than posts or comments.
Example: subreddit_name="pics", language="en-US", need_format=false.
ae_xiaohongshu_fetch_homefeed
Retrieve Xiaohongshu homepage recommended notes, optionally by category channel and with image-and-text notes only.
Inputs (all optional):
- num: Number of notes requested; defaults to 20, maximum 40.
- category: Category channel ID; defaults to "homefeed_recommend". Use a channel ID, not a search keyword.
- cursor_score: Pagination cursor; leave empty for the first page. Defaults to "".
- need_filter_image: Set to true for image-and-text notes only, or false for mixed recommendations including videos. Defaults to false.
Returns response data containing the recommended notes. The available schema does not specify individual note fields or pagination response fields.
ae_xiaohongshu_search_groups
Search Xiaohongshu group chats by keyword, with support for pagination.
Inputs:
- keyword (required): Search term, such as "上海".
- page_no (optional, default 0): Page number, starting at 0.
- search_id (optional, default ""): Leave empty for the first request; it is generated automatically. For later pages, reuse the search_id returned by the first search.
- source (optional, default "unifiedSearchGroup"): Search source.
- is_recommend (optional, default 0): Recommendation flag; 0 means no and 1 means yes.
Returns: Search result data containing a group list and pagination information. Individual group record fields are not specified.
To paginate, start with page_no=0 and an empty search_id, then increment page_no while reusing the first search's search_id. This tool finds groups; it does not join them or read their chat messages.
ae_pipixia_get_home_short_drama_feed
Retrieve home-page short-drama recommendations.
Input: page is an optional integer with a default of 1. Start with page=1 and increase it by 1 for each subsequent page.
Returns: a response containing the requested page's short-drama recommendation data in data. The payload may be null, and individual feed item fields are not specified.
ae_pipixia_search
Search Pipixia for posts, users, and topics by keyword.
Inputs:
- keyword (required, string): The search keyword.
- search_type (optional, string; default "1"): "1" comprehensive, "8" hot, "9" fresh, "2" video, "3" photo/text posts, "4" users, or "5" topics/hashtags.
- offset (optional, string; default "0"): Pagination cursor. Start with "0"; for subsequent pages, use the offset value returned by the previous response, not a page number.
Returns: A search response whose data contains the search results. The detailed result structure is not specified and may vary by search type.
Example: keyword="皮皮虾", search_type="1", offset="0".
ae_lemon8_search
Search Lemon8 for content, accounts, or hashtags by keyword.
Inputs:
- query (required string): Search keyword.
- search_tab (optional string, default "main"): "main" for Top/comprehensive search, "user" for Accounts, "hashtag" for Hashtags, or "article" for Posts.
- filter_type (optional string, default ""): "" for all content, "video" for videos, or "posts" for posts.
- order_by (optional string, default ""): "" for relevance, "popular" for popularity, or "recent" for newest first.
- max_cursor (optional string, default ""): Leave empty for the first page. For later pages, use the previous response's max_cursor with the same search criteria. Check has_more to determine whether more results are available.
Returns: The requested page of search results in the response's data field, with pagination information used to continue the search. Detailed result-item fields are not specified.
ae_threads_search_posts
Search Threads posts by keyword.
Input:
- `query` (required string): The keyword to search for, such as `bitcoin`.
Returns:
Search response data with a documented `results` list of matching posts. Entries may include post details such as identifiers, captions, like counts, and timestamps; author information such as usernames, profile images, and verification status; and image, video, or carousel media information when available. The detailed result structure is not fixed by the response schema, so available fields may vary.
Example: `{"query":"bitcoin"}`
ae_linkedin_get_company_posts
Retrieve posts from a LinkedIn company page.
Inputs:
- `url` (required): The LinkedIn company page URL, for example `https://www.linkedin.com/company/microsoft/`.
- `start` (optional): Nonnegative pagination offset. Use 0 for the first page, 50 for the second, and continue in increments of 50.
- `sort_by` (optional): `top` for top posts or `recent` for latest posts.
- `pagination_token` (optional): Unsupported and has no effect. Use `start` to paginate.
Returns company post data and paging information (`count`, `start`, and `pagination_token`). A returned pagination token cannot be used to fetch another page.
Temporary 400 or other errors may occur under load. Automatic retries are recommended, with a retry count of 3.
ae_xigua_search_video
Search Xigua Video by keyword, with optional sorting, duration filters, and pagination.
Inputs:
- keyword (required string): The search keyword.
- offset (optional integer, default 0): Use 0 for the first request. For later pages, use the offset returned by the previous response.
- order_type (optional string): Omit or leave empty for default ordering. Use publish_time for newest videos or play_count for most played.
- min_duration (optional integer): Minimum video duration in seconds; omit if not needed.
- max_duration (optional integer): Maximum video duration in seconds; omit if not needed.
Returns: A response containing the video search results. Detailed video item fields are not specified. Use the returned offset for subsequent pages when available.
Example: keyword="travel", order_type="play_count", min_duration=60, max_duration=180, offset=0 searches for travel videos lasting 1–3 minutes, sorted by play count.