Coevera

Exports Coevera CRM records to a CSV file and returns a one-time download link instead of the records. The response holds no rows, only the link and a description of the file. search_records and get_records return the records themselves, up to 100 per call; run_report computes counts, sums and breakdowns in the CRM without a file. The link: • works for one download: the first download, by anyone, uses it up; a second request gets HTTP 410; • needs no login: whoever has it can download the file until it is used or expires (expires_in_seconds); • can be opened in a browser or downloaded by code (curl, Python). The file is not stored on the server; it is read from the CRM during the download. Each call creates a new, independent link. Response shape: • download_url, expires_in_seconds — null when nothing matches • row_count — rows in the file; total_matching — records that match the filter • truncated, truncated_reason — the file has fewer rows than total_matching: "max_rows" (only the first max_rows records), "crm_error" (the CRM failed to load a record and the file stops before it; see crm_error, failed_after_id), "time_budget" / "crm_rate_limit" (counting stopped early) • file — how the file is written: encoding, delimiter, decimal separator, date formats, booleans, null • columns — one entry per column in file order: name (the header), type, content, references, options, currency • inaccessible_fields — fields left out because the user's role cannot read them Values: headers are field api_names, composite parts are dotted (value.base_value). References (owner, step, types, currency), relations (primary_account, primary_contact) and dropdowns hold names and labels, not ids, also in *_id columns; with include_ids, each of them is followed by an id column. Date-times are UTC. Only records and fields the user may see are exported, and exporting requires the Export permission of the user's CRM role.

export_records

Fetches records by ID from a Coevera CRM entity. Returns {entity, count, records} where records is the full list of matching records in the same order as record_ids (records not found in the CRM are simply absent). fields: list of field api_names to include in each record. expand (optional): list of field api_names to expand as nested objects. Only supported for relation fields. Invoke when specific record IDs are already known and their field values are needed.

get_records

Returns the checklist items configured on a pipeline step — the requirements that must be satisfied before an opportunity can advance to that step. step_id: UUID of the pipeline step to inspect. opportunity_id: Optional UUID of a specific opportunity. When provided, each item is annotated with is_complete (True/False) showing whether that requirement is already satisfied for that opportunity. When omitted, only the definitions are returned. item_type values: FieldType — a specific CRM field must have a value (field_id identifies which) ActivityType — a task or appointment of a specific type must be linked or completed (activity_type_id identifies the type; item_validation specifies whether linking or completing is required) CreateNote — a note must be created on the record AttachDocument — a file must be attached UseDocumentTemplate — a document template must be filled and attached SendEmail — an email must be sent from the record EnrollToEmailSequence — record must be enrolled in an email sequence RunProcess — a process automation must be triggered Default — generic manual requirement item_validation values (relevant for ActivityType items): Required — the activity must exist RelatedActivityLinked — a matching activity must be linked to the record RelatedActivityCompleted — a matching activity must be completed Results are ordered by sort_order ascending.

get_step_checklist_status

Searches Coevera CRM records of several entity types at once for a text, like the CRM's search bar: "find Acme", "everything about John Smith", "which record is +1 555 0100". No field names are needed. The best matches come first. Matching: case- and accent-insensitive. A record matches when EVERY word appears somewhere in its searched fields (as a substring; words may be in different fields), OR when the whole text is a close fuzzy match for one field, so a multi-word query can also return near misses. Words of 1–2 characters only match the start of a word. A phrase in double quotes must appear exactly as written and is never matched fuzzily: quote names, emails and phone numbers for precise results. Searched entities: Account, Contact, Lead, Opportunity, Task, Appointment, Project, CloudObject and custom entities with global search enabled (see global_search in meta_list_entity_types; meta_describe_entity lists the fields searched). Only records the user may see are returned. A record can match on any of its searched fields, including ones not shown in its card (for example a custom field), so the card does not always show why it matched. Response shape: • counts — {<entity>: number of matches} for entities with at least one match • total — sum of counts • results — [{entity, id, name, summary, archived, url}], a compact card per record • next_cursor — pass as cursor, with the same query and entity_names, to get the next page; null on the last page. counts and total always cover all pages. Use get_records with entity and id for full details. Use search_records instead when filtering by field values (status, dates, owner) or for entities not listed above.

global_search

Returns all entity types available in the Coevera CRM for the authenticated user. Each entry includes the entity name, display label, kind (primary or relation), api_name, custom_name, description, and the user's CRUD permissions (read, update, write, delete as booleans). global_search.supported tells whether the global_search tool searches the entity. Invoke when the entity name is unknown or when available entities and permissions need to be enumerated.

meta_list_entity_types

Aggregates Coevera CRM records in a single call: counts, sums, averages, minimum and maximum, optionally grouped by up to 3 fields. Use it for totals and breakdowns such as "open pipeline value by step and owner", "weighted forecast", "deals closing per quarter" or "leads by source" instead of paging through search_records and adding values up. Response shape: • groups — [{key: {<group field>: value}, metrics: {<metric>: number}}], largest first • totals — the same metrics over all scanned records • group_count / groups_truncated — number of groups and whether only the top limit_groups are returned • scanned / total_matching / truncated / truncated_reason / scan_cap — how many records were read. When truncated is true, only the first `scanned` matching records were aggregated: truncated_reason "scan_cap" (the cap was reached), "time_budget" (the call ran out of time) or "crm_rate_limit" (the CRM refused more requests). Narrow the filter for exact totals, and say the result is partial • base_currency — currency of all currency amounts (present when a currency field is aggregated) Group keys for users, steps, sales units, currencies and dropdown fields are returned as {id, name}. Date fields can be grouped by period with ':month', ':quarter' or ':year' (e.g. "closing_date:quarter"); periods use the stored date (UTC for date-times). Currency fields (e.g. Opportunity value) are aggregated in the space's base currency. Group by '<currency field>.currency_id' (e.g. "value.currency_id") for a per-currency breakdown. weighted=true (Opportunity only) adds weighted_value = value × probability of the deal's current step, counted for open opportunities only (won and lost deals add nothing). To report on open pipeline only, filter by status, e.g. {"status": {"eq": ["1"]}} (Opportunity status: 1 Open, 2 Lost, 3 Won). Only records the user is allowed to see are included. Use meta_describe_entity first to find field api_names.

run_report

Searches a Coevera CRM entity and returns all matching records for the current page. Response shape: • entity — entity name • count — number of records returned in this page • records — list of all normalized matching records • pagination — {has_next_page: bool, next_cursor: str|null} To fetch the next page, call again with the same arguments and pass pagination.next_cursor as `after` (`cursor` is accepted too). Unknown argument names are rejected. fields: list of field api_names to include in each record. filter (optional): boolean filter tree — see schema for structure. Omit to match all records. order_by (optional): list of sort objects mapping a field name to 'Asc' or 'Desc'. limit: number of records to return per page (default 100, max 100). after (optional): pagination.next_cursor from the previous response. Omit for the first page. cursor (optional): the same as after, accepted under this name too (global_search uses it). expand (optional): list of field api_names to expand as nested objects. Only supported for relation fields. Invoke when records matching specific field criteria need to be found.

search_records

Sets the picture of an Account (logo), Contact (photo) or Product (product image), replacing the current picture. Two phases: this call checks the record and returns a one-time upload_url; then POST the image to it as multipart/form-data, e.g. curl -X POST {upload_url} -F file=@/path/to/logo.png. Accepted formats: PNG, JPEG, GIF, WebP; the file content is checked, not only its name. Check the file size against max_file_bytes BEFORE uploading; a larger file is rejected with HTTP 413 and the one-time upload_url is then spent. The upload endpoint responds with entity_name, record_id, record_url and the stored picture (id, filename, mime_type, size, width, height, url). Files uploaded with upload_cloud_object_document are documents and cannot be used as a picture.

set_record_picture

Partially updates existing records in a Coevera CRM entity; only the fields provided in each record dict are changed. Each record dict must include the record's "id" field to identify the target. Returns a result list in the same order as the input: each entry is either {"id": "...", "url": "..."} on success or an error object on failure. Records with missing id or read-only field violations are skipped and reported as errors; valid records are still submitted. Invoke to modify one or more existing records.

update_records

Initiates a two-phase file upload to Coevera CRM. Returns upload_url, upload_token, cloud_object_id, expires_in_seconds, max_file_bytes and max_file_mib. Check the file size against max_file_bytes BEFORE uploading — larger files are rejected with HTTP 413 and the one-time upload_url is then spent. After receiving these, upload the file binary to upload_url via: curl -X POST {upload_url} -F file=@/path/to/file The upload endpoint responds with the stored file nested under createCloudObject.cloudObject (id, filename, mimeType, size, url, publicUrl). The file is stored as a document; it cannot be used as the picture or logo of a record (use set_record_picture for that).

upload_cloud_object_document