Ragic

Full-text search across ALL sheets in one database, for when you do NOT know which sheet the data is in. Provide content keywords as `query` (plain words, NOT field:value syntax). Returns ranked hits, each with sheet_id and record_id that you can pass directly to get_records, plus a snippet and record_url. Use this as the entry point when the user describes data in natural language without knowing its location; once you have a hit's sheet_id you can also call list_sheets / list_page to refine. Searches a single database only; to cover multiple databases, call this once per database.

full_text_search

Retrieve one or more FULL records from a sheet, each including subtables, a `comments` array (record-level discussion: user_email, user_name, timestamp, content) plus `comment_count`, a `cell_comments` array (field-level notes added via the UI right-click "Insert Comment": field_id, field_label, cell_node_id, user_email, user_name, timestamp, content) plus `cell_comment_count`, and an `email_conversation` array (emails sent/received on this record: timestamp, from, to, cc, subject, content as plain text with quoted replies stripped, attachments) plus `email_conversation_count`. Choose records in ONE of two ways: (a) pass `record_id` as a single rootNodeId integer or an array of rootNodeIds to fetch those records directly; or (b) omit `record_id` and pass `filters` to query. `count` caps how many records a filter query returns (default 50, max 1000). For a lightweight listing-page view (only listing columns, more rows) use list_page instead. Each filter entry can be either a plain string (matched with like/LK as a shorthand), or an object {op, value} for a typed comparison. Supported op values: "=" (eq), ">=" (gte), "<=" (lte), ">" (gt), "<" (lt), "like" (default for strings), "range" (uses hi/low instead of value, inclusive both ends), "regex", "fts" (full-text search). Filter queries are paginated with `count` as the page size: if the response includes next_cursor, pass it back as `cursor` to fetch the next page (pagination does not apply when record_id is given). When the user asks for a large amount of data or ALL matching records, call this tool repeatedly, passing each response's next_cursor as `cursor`, until the response no longer includes next_cursor.

get_records

Browse records as they appear on a sheet's listing page: only the listing-page columns are returned (a curated subset of fields, in listing order), which is lightweight for scanning or locating records. Returns a `fields` array describing those columns and a `records` array, each entry carrying its record_id. `filters` is OPTIONAL: omit it to list the sheet's records, or provide a map keyed by field_id (from list_sheets) or by the special record-metadata keys _create_date / _create_user / _entry_manager to filter. Each filter entry can be either a plain string (matched with like/LK as a shorthand), or an object {op, value} for a typed comparison. Supported op values: "=" (eq), ">=" (gte), "<=" (lte), ">" (gt), "<" (lt), "like" (default for strings), "range" (uses hi/low instead of value, inclusive both ends), "regex", "fts" (full-text search). Records are returned as an ordered array (each with record_id), following the listing page's sort order. `limit` is the page size (default 10, max 50). Results are paginated: if the response includes next_cursor, pass it back as `cursor` to fetch the next page. For the FULL content of specific records (subtables + comments) use get_records.

list_page

List all Ragic databases this connector can access. Call this first: pass a returned database id as the database argument of the other tools. Cache the result until cache_ttl_seconds expires.

list_databases

List all sheets (forms) in the given database, along with each sheet's fields (field_id, label, type). Call this after list_databases to discover available sheets and their schema before calling list_page or get_records.

list_sheets

List user candidates for one record-metadata filter key, using the same option source as the Ragic list page sidebar for that system filter. This is not a database address book. `filter_key` is REQUIRED and must be `_create_user` (who created a record) or `_entry_manager` (a user assigned as a record manager). Use the returned `email` value in list_page / get_records filters. `query` is optional: pass a name or e-mail fragment to search (case-insensitive, matched against both e-mail and name), or omit it to list candidates from that sidebar-equivalent option source. The candidate source is per database + filter_key; sheet_id is not required because these system-field options are not narrowed per sheet in the current list page sidebar implementation. Each user is returned as {email, name}. Only active user values that appear in the filter option source are returned; this can differ from all users in the database or users who share groups with you. Paginated: `limit` is the page size (default 50, max 200); if the response includes next_cursor, pass it back as `cursor` to fetch the next page.

list_users

Read Ragic's developer documentation. Use this to answer questions about Ragic's HTTP API (integrating external systems, reading/writing records over HTTP) or the JavaScript Workflow scripting engine (writing workflow scripts, action buttons, approval flows). This tool takes no database; the documentation is global and read directly on this server. doc=http_api: page is "cheatsheet" (default, an LLM-oriented summary of the whole HTTP API; prefer this) or "openapi" (the full OpenAPI spec in YAML, very large - only fetch it when the cheatsheet is not detailed enough). doc=workflow: omit page to get the list of available page ids plus a table of contents, then call again with a page id (e.g. "01-getting-started", "api/entry", "recipes/external-api") to read that page's markdown. lang is "en" (default) or "zh-TW"; any other value falls back to en.

read_documentation