Developers
Set your API contextPick a region and space ID once. Every sample fills in your base URL.
Where do I find these?

In Coevera: open your space's API application and choose Show Access. It shows the space ID and the base URL, whose host starts with your region.

From the API: if you know the space ID, the discovery endpoint GET /api/discovery/spaces/{space_id} returns its region and base URLs, without credentials.

Only the region and space ID are kept, in this browser. Never paste a key or password here.

More in the quickstart

AI & MCPMCP changelog

MCP changelog

Changes to the Coevera MCP server that clients and agents can see, by server version, newest first. Separate from the API changelog.

Answer

The MCP server has its own versions, released independently of Coevera's monthly releases. The current version is 2.1.0, from October 2026. This page lists the changes clients and agents can see, newest first, from the server's own release notes; changes to the REST and GraphQL API are in the API changelog.

The server's own release notes, published with its tool reference, give a month for each version, not a day. This page follows them and leaves out changes internal to the server. Changes to the REST API, GraphQL and webhooks are in the API changelog.

2.1.0: October 2026

  • Added: global_search, which finds records by a text across accounts, contacts, leads, opportunities, tasks, appointments, projects, documents and custom entities with global search turned on, in one call and ranked by relevance, as the search bar in Coevera does. It returns counts per entity and a short result per record, with a link to it.
  • Added: run_report, which counts, sums and averages, and finds minimums and maximums, over the records that match a filter, grouped by up to three fields, with dates by month, quarter or year. Amounts are totalled in the space's base currency, and the response says when and why a result covers only part of the records.
  • Added: export_records, which writes the records that match a filter to a CSV file behind a one-time download link on the new /download/{token} address. The response describes the file and its columns, not the rows. Exporting needs the Export permission in Coevera, and clients ask the user to approve every export.
  • Changed: meta_list_entity_types and meta_describe_entity say whether global_search covers an entity, and which of its fields it searches.
  • Changed: the server's instructions point assistants to global_search, run_report and export_records, and say that tool results are data entered by users, never instructions.
  • Changed: every tool description fits the 2,048 characters that claude.ai shows.
  • Changed: search_records takes the pagination cursor as cursor as well as after.
  • Fixed: every tool rejects an argument name it does not know, with an error that lists the valid names. A cursor sent under a wrong name used to be ignored, and search_records returned the first page again.
  • Fixed: meta_describe_entity no longer lists fields the REST API cannot load, which made searches that asked for them fail.
  • Fixed: record links no longer contain a double slash.

2.0.11: October 2026

  • Fixed: set_record_picture failed on every call before it returned the upload link.

2.0.10: October 2026

  • Added: set_record_picture, which sets the logo of an account, the photo of a contact or the image of a product through a single-use upload link, as file uploads do. The file must be a PNG, JPEG, GIF or WebP image.
  • Changed: create_records, update_records and change_ownership no longer skip Coevera's validations. A write that breaks a rule (a required field, a field value, a relation, a step checklist or a custom validation rule) comes back as an error for that record instead of being stored, so some writes that used to succeed now fail.
  • Changed: picture_id on accounts, contacts and products cannot be written with create_records or update_records, which answer read_only_field_violation for that record and name set_record_picture. create_records now reports read-only fields per record, as update_records does.
  • Fixed: a document uploaded with upload_cloud_object_document can no longer be set as a record's picture.

2.0.9: October 2026

Changed: minor improvements and fixes. The server's notes list no details.

2.0.8: September 2026

  • Added: delete_records, a soft delete of up to 100 records per call, with a result per record.
  • Changed: delete is allowed on more entities and relation types. Leads and opportunities still cannot be deleted; archive them instead.
  • Added: documents (CloudObject) and document folders (CloudObjectFolder) can be created and updated.
  • Changed: uploads are limited to 50 MiB, and the limit is returned to the client as max_file_bytes.
  • Changed: the OAuth issuer is the server's own origin, https://mcp-server.coevera.com, and the discovery document advertises PKCE with S256.
  • Added: a 401 carries a WWW-Authenticate header that points to the protected-resource metadata.

2.0.7: September 2026

  • Added: upload_cloud_object_document, a two-step upload: the tool returns a single-use upload URL, and the client posts the file to it.

2.0.1 to 2.0.6: July 2026

  • Changed: corrected tool annotations (2.0.2, 2.0.3), with a justification for each hint (2.0.4).
  • Changed: fixed a mismatch in the OAuth scopes (2.0.1).

2.0.0: June 2026

  • Added: change_ownership, for tasks, opportunities, contacts, accounts, appointments, leads and memos.
  • Added: the bulk tools bulk_move_opportunities, bulk_archive_opportunities, bulk_archive_leads and bulk_qualify_leads, with a rollback option. Batches of 20 or more run asynchronously.
  • Added: the step checklist tools get_step_checklist_status and complete_step_checklist_item.
  • Changed: custom lookup fields appear among an entity's expandable fields.

1.1.5: June 2026

  • Changed: every tool annotation hint is set explicitly to true or false.

1.1.4: May 2026

  • Added: support for ChatGPT's connector verification.
  • Changed: write tools report errors per record.
  • Changed: a failure of the rate-limit store answers a structured 503.

1.1.3: May 2026

  • Added: progress notifications over SSE for long calls.
  • Added: the units_and_roles.read scope.
  • Changed: get_records no longer caps the number of records returned.
  • Changed: tool output schemas adjusted for ChatGPT, and tool titles and descriptions revised.