Userback

Create a new manual feedback item in a project. Use when creating feedback from scratch with project_id and description. Supports optional title/type/assignee/priority/due date. Write clear, descriptive feedback so teammates can act on it.

add_feedback

Add a comment or reply to a feedback item. Comments may be public or internal to the account - prefer internal unless specifically replying publicly to a customer. Requires a feedback_id; Use search tools first if the ID is unknown.

add_feedback_comment

Load full details for known feedback IDs, including screenshot urls and comments. Do not use as a search tool, and do not call it for every search result. Use search tools for lists and call get_feedback only for selected IDs. Use get_feedback_logs when you only need technical console/network diagnostics. Render one section per feedback with key fields, short summary, and notable comments. Always include feedback_app_url for each feedback item so clients can deep-link directly.For multiple feedback items, include a compact overview first, then detailed sections.

get_feedback

Retrieve the most recent console and network log entries for a specific feedback item. Does not support pagination. Use for troubleshooting a specific feedback. Use get_feedback for screenshots, comments, and general feedback details. Returns newest entries first for both console_logs and network_logs. Use limit to cap results per log type; defaults to 20 and maxes at 100.

get_feedback_logs

List account members, optionally scoped to a project and optionally filtered to assignable members. Use to discover user IDs, roles, and names before assigning feedback. Optionally set project_id to return only members with specific project access. Optionally set assignable_only=true to filter to assignable roles.

list_account_users

List accessible projects. When to use: discover accessible projects or fetch workflow mappings for UI choices. Set include_workflows=true when workflow names or IDs may be needed next, avoiding extra list_workflows calls per project. Do not call before every search; search tools already enforce project access.

list_projects

Optional discovery helper: list workflow states for one project in sort order. Workflows are project-scoped; no global states exist. Use when explicitly enumerating states for UI selection. Workflow names and IDs are only valid within the given project_id. Do not call as a preflight on every request - to update the workflow status for a feedback item call set_feedback_workflow with workflow_name or workflow_id.

list_workflows

Deterministic feedback search with structured filters, substring exact matching, sorting, counts, and pagination. Search choice for lists, totals, broad retrieval, or ambiguous tool choice. Use this tool when reproducible coverage and page-by-page control are needed. The search argument is one exact substring matched against title/description, not AND/OR syntax. Use search_feedback_semantic for intent or meaning searches. Start with a small limit and first page, summarize count, then ask before iterating additional pages. Do not auto-loop pages to exhaustion unless the user explicitly requests full export, bulk retrieval, or all feedback. Avoid first-call all-project scans with max limits when a narrower first pass can validate intent. Do not call get_feedback after searching unless selected items need details not returned by search. Issue type is provided as feedback_type (bug, idea, feedback). Render filtered results with a compact summary of count, filters, and sort followed by item sections. When referencing returned feedback, include `feedback_app_url` so clients can deep-link to the item.

search_feedback_filter

Semantic relevance retrieval using vector search. Use for meaning-based discovery when exact keywords or filters are unknown and ranking by relevance is desired. Results are relevance-ranked and may be non-exhaustive. Use search_feedback_filter for deterministic filters, counts, paging, stable ordering, or exhaustive coverage. Keep query specific enough to avoid broad noisy matches. One topic per search. Keep the search query simple, do not overexpand and complicate the query. Start with a smaller limit, inspect top matches, then refine the query or confirm expansion. Do not set min_score unless specifically requested. Do not treat semantic output as complete inventory unless the user explicitly accepts approximate ranked retrieval. If you already have IDs, use get_feedback directly instead of searching again. Do not call get_feedback after searching unless selected items need details not returned by search. Render results grouped by best matches first with concise rationale per item, and include feedback_app_url.

search_feedback_semantic

Assign or unassign a known feedback item to a project member. Use list_account_users first when valid user IDs are needed. Set assignee_user_id to 0 to unassign.

set_feedback_assignee