Page through large results

Every list tool that can return an unbounded number of rows is paginated the same way: ask for a page size with first, and walk forward with after.

Read the first page

list_pages(project_id, first: 20)

The response carries the rows plus two things you need for the next call:

{
  "totalCount": 214,
  "pageInfo": { "hasNextPage": true, "endCursor": "YXJyYXljb25uZ…" },
  "pages": [ … ]
}

Read the next one

Pass the previous response’s endCursor back as after:

list_pages(project_id, first: 20, after: "YXJyYXljb25uZ…")

Repeat while pageInfo.hasNextPage is true.

The rules

• after is an opaque cursor, not a page number and not an offset. It only means “continue after this row”. Do not construct one, do not increment one.
• totalCount is the size of the whole result set, not of the page you got.
• There is no backward pagination. Forward only.

Which tools paginate

Tool	Default first
list_projects	20
list_pages	20
list_project_run_history	20
list_agentic_tickets	20 (max 100)

list_journeys and list_journey_groups return the project’s full set.

Run history gets long fast

list_project_run_history includes every scan your CI reports through b8e ci, which is one entry per push. On an active repository that is thousands of entries. Take the first page and only walk back if the answer is not in it.