Rovo’s built-in user and issue search tools cap at 50 or 200 results. That breaks access reviews and backlog analysis. Here is the pattern we use to get around it.
The problem:
jira_atlassian_user_search_by_query returns max 50 users. No pagination. For groups with 500 people, your agent only sees 10 percent of the data. Analysis becomes random.
The fix:
Move the heavy lifting to a Forge Action. Forge can call Admin and Jira APIs with full pagination. It pulls all 500 users or 2,000 issues, does the counting or filtering in code, then returns a small JSON summary to Rovo.
Example flow:
Rovo receives: Audit the ENG-Admins group for inactive users
Rovo calls Forge Action get-group-members-full with groupId
Forge pages through /admin/v1/orgs/.../members and /audit endpoints
Forge returns: total 512 users, 87 inactive over 180 days, list of accountIds
Rovo writes the report and creates Jira tickets for the 87 users
Result:
Rovo stays fast and deterministic because it only sees summary stats. Forge handles scale. We use this for access reviews, backlog health, and license audits. No more sampling or truncated tables.
HEMANT SAINI
0 comments