conversation_continue
ChatGPTQueues a message in an existing conversation in an organization the authenticated user can access. Returns the nullable absolute conversation URL and a sanitized submission receipt with delivery presentation; the receipt does not establish that the message was delivered or read.
conversation_get
ChatGPTGets one conversation in an organization the authenticated user can access. Returns its nullable absolute conversation URL, approved sanitized message text, validated flow-completion references, fixed delivery presentation, sanitized current-user inbox context when present, and a reply action hint.
conversation_start
ChatGPTStarts a new conversation from a response and queues the first message. Returns nullable absolute URLs for the conversation and source response, plus a sanitized submission receipt with delivery presentation; the receipt does not establish that the message was delivered or read.
event_count
ChatGPTAnswers one occurrence-count question for an authorized organization with Analytics enabled. Supports one named catalog event or all recorded events over rolling 7/30/90 elapsed days or current workspace calendar month to date. Conditions, property filters, breakdowns, people counts, arbitrary dates and SQL are unsupported. Returns only organization-level total/daily counts, executed scope including trusted workspace timezone and unverified freshness. Named selection is unavailable when the bounded catalog overflows. No raw records or identities; counts do not claim anonymity.
feedback_report
ChatGPTThis reports feedback about getuserfeedback.com. The source organization is used only for authorization, is not sent as metadata, and no account identity or history is added automatically. Send one comment-only product feedback report. For user_relayed, use only a real user statement or a summary the user approved. For agent_observation, describe a direct observation. Include only bounded, relevant context in properties; this tool does not collect supporting evidence or promise resolution.
flow_activity_get
ChatGPTGet a flow’s delivered, viewed, and responded counts by delivery batch or calendar week.
flow_content_insert
ChatGPTInserts one Question or Message block as a new Page in an already canonical questionnaire. Use an explicit destination and anchor from flow_get.flow.insertionTargets with its currentVersionId as expectedVersionId; stale requests are rejected. The block _key becomes its Page key and must not collide with any existing Page or block. Existing root, rule and Otherwise sequences are supported; this does not create branches, move/remove Pages or convert legacy flows. Returns a fresh version receipt for repeated additions without echoing content.
flow_content_update
ChatGPTReplaces one flow's current page content. On canonical questionnaire flows this is content-only: retain existing Pages and routing, and use flow_content_insert for structural additions. expectedVersionId is the editableContent.versionId revision; stale revisions are rejected. For noncanonical flows, omitting navigation can replace custom routing; omitted config preserves configuration. Returns a new version receipt without echoing content.
flow_create
ChatGPTCreates a flow in an organization the authenticated user can access. Without launchSurface, the flow starts on Widget rules delivery with no live window; explicit app_delivery requires appDisplayMode imperative and starts with its live delivery window open. Returns the created flowId, organizationId, and nullable absolute flow URL.
flow_delivery_rule_upsert
ChatGPTCreates or updates the standalone delivery rule for a flow. An edit replaces the complete current rule; stale edits are rejected using the automation version ID returned by flow_get. The rule uses the canonical trigger, targeting, frequency, and delay contract. When communication purposes are enabled, communicationBudgetId must identify an active purpose. Existing journey-managed or legacy multi-rule delivery cannot be edited by this tool. Successful results include a nullable absolute flow url.
flow_get
ChatGPTGets one flow in an organization the authenticated user can access. Returns the nullable canonical absolute app URL, summary fields, launchPlan, flowSchedule delivery/window evidence, and closed, sanitized flow content. Contact comparison literals and authored routes are withheld; currentVersion.contentWithheld reports omissions and editableContent is null whenever content cannot safely round-trip.
flow_launch_surface_update
ChatGPTSelects where a flow can launch and its app display mode. Leaving Widget rules delivery may stop outstanding standalone deliveries, so the first call returns confirmation_required with a revision; explicitly confirm and resubmit that exact revision to proceed. Successful results include the persisted launch settings and current launch state, plus a nullable absolute flow url when the flow exists.
flow_publish
ChatGPTPublishes a flow on its public web page. If publishing would stop existing Widget rules deliveries, the first call returns confirmation_required with a revision; explicitly confirm and retry with that exact confirmedRulesDeliveryRevision. Successful results include the persisted launch state, editor URL, and credential-free public page URL.
flow_templates_list
ChatGPTLists flow templates by lifecycle stage or free-text query, including suggested questions and timing. Returns the effective lifecycle stage (null means all stages) and whether a nonblank query was applied.
flows_list
ChatGPTLists a compact page of flows in an organization the authenticated user can access. Returns the effective filters, including archived (current by default) and publication (null means unrestricted), whether a nonblank name or ID query was applied, and the effective limit (1–50; 10 by default). Query text is never returned. Each flow includes its nullable canonical absolute app URL.
identity_resolution_config_edit
ChatGPTApplies an editor event using the same model as the web dialog. This changes only the supplied editor state; it does not save or apply identity rules. Submit produces the shared canonical save request in state.request.
identity_resolution_config_get
ChatGPTReturns the shared identity resolution editor model for an organization administrator, opening the pending editable preview when one exists. Saving an edited preview restarts calculation and retains the old preview history. Keep its opening configuration revision when saving. No respondent values or integration credentials are returned.
identity_resolution_config_save
ChatGPTSaves using the editor's original configuration revision. Without reprocessing, applies rules to subsequent processing without rebuilding existing profiles and returns applied status. With reprocessing, saves a preview without applying it and identifies the rollout in building_preview status. Stale settings and conflicting pending previews are rejected before writes. Do not retry an uncertain save until checking current settings.
inbox_item_archive
ChatGPTArchives one conversation inbox item for the authenticated user in an organization they can access.
inbox_item_mark_read
ChatGPTMarks one conversation inbox item as read for the authenticated user in an organization they can access.
inbox_item_mark_unread
ChatGPTMarks one conversation inbox item as unread for the authenticated user in an organization they can access.
inbox_item_star
ChatGPTStars one conversation inbox item for the authenticated user in an organization they can access. Pass the inboxItemId from conversation_get, response_get, or inbox_items_list, not a conversation or response ID. A didUpdate=false result does not by itself establish the item's current state.
inbox_item_unarchive
ChatGPTUnarchives one conversation inbox item for the authenticated user in an organization they can access.
inbox_item_unstar
ChatGPTUnstars one conversation inbox item for the authenticated user in an organization they can access. Pass the inboxItemId from conversation_get, response_get, or inbox_items_list, not a conversation or response ID. A didUpdate=false result does not by itself establish the item's current state.
inbox_items_list
ChatGPTLists conversation inbox items for the authenticated user in an organization they can access. Returns the effective status and limit (10 by default; maximum 100). Status all includes non-archived items, unread includes unread non-archived items, and archived includes archived items. Pages use the inbox's priority order and may repeat or skip items if the inbox changes between calls. Returns sanitized titles and preview text, nullable absolute URLs for supported conversation or response sources, and approved record references and action hints. URLs are null when unavailable, including unknown or feature-disabled destinations. The inboxItemId identifies the inbox row, not the conversation or response.
inbox_items_mark_all_read
ChatGPTMarks all non-archived conversation inbox items as read for the authenticated user in an organization they can access.
organizations_list
ChatGPTLists organizations the authenticated user can access, with organizationId, name, and nullable absolute app home url. Names are not unique. An empty list immediately after sign-up can mean workspace provisioning is still in progress. Widget installation and identity verification are documented at getuserfeedback://docs/get-started.
pending_sends_cancel
ChatGPTCancels not-yet-delivered flow sends for a target (one flow, or the whole organization) in an organization the authenticated user can access. Only sends not yet shown to anyone are targeted; a send being claimed for delivery at the same moment may still be stopped before it is shown. apply defaults to false, which reports the current pending count without canceling; true applies the cancellation. The cancel can be narrowed to a time window with optional inclusive pendingSince and pendingUntil queued-time bounds; omitting both selects all pending sends for the target. pending_sends_restore lifts all cancellations for the target at once, not one window. Returns the nullable flow URL or organization home URL for the resolved target.
pending_sends_restore
ChatGPTLifts every cancellation pending_sends_cancel placed for a target (one flow, or the whole organization) in an organization the authenticated user can access, all at once — including windowed cancels. Other holds are unaffected: an archived or switched-off flow, revoked access, and a cancellation at the other scope (restoring a flow does not lift an organization-wide cancel) stay in place; a paused flow still holds back its automatic sends, but not "send now" sends. Passing pendingSince or pendingUntil is an error. Returns how many cancellations were lifted and the nullable flow URL or organization home URL for the target — a send resumes once nothing else holds it back. Idempotent: already-lifted cancellations are left unchanged.
product_knowledge_get
ChatGPTRead a public source by document ID returned by product_knowledge_search, optionally restricted to a section. Unknown IDs return null. No organizationId is required.
product_knowledge_search
ChatGPTSearch public product docs, pricing, templates, blog and question library. Returns relevance-ordered source passages with canonical citations and revisions. Relevance does not establish answer correctness. Use product_knowledge_get to read a source. No organizationId is required. Use flow_templates_list for actionable template IDs.
response_get
ChatGPTGets one response in an organization the authenticated user can access. Returns nullable absolute response and flow urls, a nullable respondentUrl when the opaque respondentReference identifies a usable current profile, approved end-user-provided answer content, sanitized analysis and flow text, conversations that mention the response, and conversation action hints. Explicit identities, metadata, events, contact-answer fields, and unknown answer shapes are withheld.
responses_list
ChatGPTLists a compact page of sanitized response summaries in an organization the authenticated user can access. Returns effective filters, including whether the flow selector, retained delivery Group context, and nonblank query were applied, and the effective limit (10 by default; maximum 100), in stable ID order. Each response includes nullable absolute response and flow urls and a nullable respondentUrl when its opaque respondentReference identifies a usable current profile. Query returns a page of matches in original approved answer content that remains visible after privacy filtering. Withheld contact fields and unclassified or invalid answers cannot match. Raw query and respondentIdentity values are not echoed; flowId and selectedGroup are withheld and their applied flags indicate whether each was used. Opaque respondentReference is returned as the active filter. References identify one current respondent within the organization and may be absent or change after identity deletion/reset. respondentIdentity and respondentReference are mutually exclusive.
segment_create
ChatGPTCreates a user segment in an organization the authenticated user can access from a structured filter definition and name. Configured user-connection/group conditions use the same in and not in membership semantics as other segment conditions and can be selected in supported Flow audience and condition controls. Status defaults to draft; member evaluation is asynchronous. Returns the persisted status, segment ID, version ID, and nullable absolute segment URL.
user_activity_page
ChatGPTReads up to 20 events in an authorized User Activity group and returns nextCursor when another bounded page exists. Each call rechecks membership and the current active User. Returns sanitized labels, sources and times; identities, attributes and dynamic properties are withheld. Missing or unavailable groups use the same error.
user_get
ChatGPTRead the User behind a Response using its opaque respondentReference and organizationId. Returns profile creation time and at most 20 recent Activity groups within the past 30 days. End-user identity fields, attributes and dynamic event properties are withheld. Sanitized event labels, activity sources, counts and timing are shared. Missing or unavailable profiles use the same error.
weekly_digest_get
ChatGPTGets one weekly feedback digest in an organization the authenticated user can access. Returns a sanitized cover, approved digest blocks, and Markdown rendered from the same projection. Quote attribution, note signatures, and arbitrary action payloads are withheld.
weekly_digests_list
ChatGPTReturns valid frozen snapshots from the 12 most recent closed weekly editions. Legacy and malformed snapshots are omitted; older history is not paginated. Each digest includes its period and sanitized cover headline, preview, and subject.
widget_config_get
ChatGPTGets the customer-facing widget configuration and health for an organization the authenticated user can access. Returns the public widget installation key, nullable absolute widget settings url, identity verification status and health, supported installation methods, activity guidance, and relevant recommended actions. Activity reports product events ingested for the provisioned widget app; inactive means none in the last 72 hours and does not by itself indicate a broken installation. Activity does not establish whether flows or responses exist. The publicApiKey is intended for frontend and mobile code and does not grant dashboard or MCP access. Stored issuer, JWKS URL, and claim mapping are withheld; use the widget settings URL when available to view or change them. Active organization administrators receive sample-token test recommendations when saved identity verification settings are valid.
widget_config_update
ChatGPTUpdates widget identity verification for an organization and returns its canonical health and a nullable widget settings url. Pass configurationRevision from the latest widget_config_get response to prevent overwriting newer settings. Requires an active organization administrator membership.
workspace_activity_list
ChatGPTLists canonical instrumented workspace events newest first for an authorized organization ID. Limit 1-100, default 50. Pass nextCursor unchanged for the next page. Uninstrumented history is not included; missing command history remains missing. Returns canonical events and current member display profiles, not arbitrary audit details or end-user identity attributes.
