feat(reports): align conversation agent metrics

docs/CHATWOOT_PARITY_DEVELOPMENT_PLAN.md

@@ -49,11 +49,11 @@ Hermes task landing checklist:
 
 ## Current Baseline
 
-- Current tracking checkpoint: 2026-06-07 live reports team-scope parity checkpoint, prepared as `feat(reports): filter grouped live metrics by team`.
-- Latest implementation checkpoint: this checkpoint, prepared as `feat(reports): filter grouped live metrics by team`.
-- Latest documentation/tooling checkpoint: this tracker update records P3.48 live reports grouped team-scope parity plus the committed P3.47/P3.46/P3.45/P3.44/P3.43/P3.42/P3.41/P3.40 reports baseline, lands the known Hermes plan families into this document, and keeps this file as the active follow-up plan that supersedes `.hermes/plans/*`.
+- Current tracking checkpoint: 2026-06-07 reports conversation agent-metric parity checkpoint, prepared as `feat(reports): align conversation agent metrics`.
+- Latest implementation checkpoint: this checkpoint, prepared as `feat(reports): align conversation agent metrics`.
+- Latest documentation/tooling checkpoint: this tracker update records P3.49 reports conversation agent-metric parity plus the committed P3.48/P3.47/P3.46/P3.45/P3.44/P3.43/P3.42/P3.41/P3.40 reports baseline, lands the known Hermes plan families into this document, and keeps this file as the active follow-up plan that supersedes `.hermes/plans/*`.
 - Plan landing status: complete for the current known Hermes plans and user-confirmed scope. Future work should update this file directly instead of opening a parallel tracker.
-- Worktree status at this implementation checkpoint: live reports grouped metrics now match inspected Chatwoot `Api::V2::Accounts::LiveReportsController#load_conversations` team scope behavior by honoring optional `team_id` before grouped live metrics are calculated. `grouped_conversation_metrics?group_by=assignee_id&team_id=...` now excludes conversations from other teams while preserving nil-assignee and open/unattended/unassigned payload shape. P3.47 grouped error-shape parity, P3.46 summary aggregate parity, P3.45 report bot metric timeseries count strategies, P3.44 live additive rollup parity, P3.43 rollup/backfill timezone parity, P3.42 timeseries timezone parity, P3.41 CSV/download parity, and P3.40 raw JSON/date parity for report, live-report, and summary-report endpoints remain implemented. Profile MFA, Facebook callbacks, account WhatsApp calls, onboarding, notification scoping, API v2 year-in-review, and WhatsApp calling toggle parity remain implemented from previous checkpoints. Live API/browser/enterprise smoke still needs the full PostgreSQL/Redis/Meilisearch/GoChat/Vite/Chrome stack.
+- Worktree status at this implementation checkpoint: v2 reports conversation metrics now match inspected Chatwoot `Api::V2::Accounts::ReportsController#conversations` and `V2::ReportBuilder#conversation_metrics/#agent_metrics` behavior: blank `type` returns empty `422`, `type=account` keeps raw live account metrics, and non-account requests now return paginated account-user agent objects sorted by open conversation count with `id/name/email/thumbnail/availability/metric` instead of grouped assignee buckets. P3.48 live grouped team-scope parity, P3.47 grouped error-shape parity, P3.46 summary aggregate parity, P3.45 report bot metric timeseries count strategies, P3.44 live additive rollup parity, P3.43 rollup/backfill timezone parity, P3.42 timeseries timezone parity, P3.41 CSV/download parity, and P3.40 raw JSON/date parity for report, live-report, and summary-report endpoints remain implemented. Profile MFA, Facebook callbacks, account WhatsApp calls, onboarding, notification scoping, API v2 year-in-review, and WhatsApp calling toggle parity remain implemented from previous checkpoints. Live API/browser/enterprise smoke still needs the full PostgreSQL/Redis/Meilisearch/GoChat/Vite/Chrome stack.
 - Next executable implementation checkpoint: continue Phase 2/3 drift audit for deeper report data-source/metric-builder drift or the next reused-frontend mismatch, Phase 6 placeholder audit, or B12 live smoke from fresh reference/smoke evidence.
 - `go test ./...` passes when run outside the restricted socket sandbox; focused Facebook callback handler/router/route-parity tests pass in the sandbox.
 - Route dump succeeds with `967` registered routes after profile MFA route tracking.
@@ -156,6 +156,7 @@ This table is the shortest authoritative handoff view. If an older lower section
 
 | Priority | Workstream | Current state | Next checkpoint | Commit close rule |
 | --- | --- | --- | --- | --- |
+| 0 | P3.49 reports conversation agent-metric parity | Implemented for legacy/reused v2 reports conversation clients: GoChat now follows inspected Chatwoot `Api::V2::Accounts::ReportsController#conversations`, `V2::ReportBuilder#conversation_metrics`, and `V2::ReportBuilder#agent_metrics`. Blank `type` returns empty `422`; `type=account` preserves raw account live metrics; non-account requests return paginated account-user objects with `id`, `name`, `email`, `thumbnail`, `availability`, and a `metric` object containing only `open` and `unattended`, sorted by open count descending. | Keep in Review; reopen from B12 reports smoke or fresh reference evidence for Kaminari pagination metadata assumptions, tie ordering, non-agent `type` scoping with `id`, or exact nil thumbnail/avatar behavior. | Focused Analytics handler and service tests, full `go test ./...`, and `git diff --check` must pass. No route artifacts change. |
 | 0 | P3.48 live reports grouped team-scope parity | Implemented for reused live-report grouped metrics clients: GoChat now follows inspected Chatwoot `Api::V2::Accounts::LiveReportsController#load_conversations`, which applies optional `team_id` before both `conversation_metrics` and `grouped_conversation_metrics`. The grouped path now parses `team_id` and filters persisted conversations before grouping by `team_id` or `assignee_id`, preserving raw open/unattended/unassigned counts and nil-assignee buckets. | Keep in Review; reopen from B12 live-report smoke or fresh reference evidence for missing-team `404` body shape, authorization status drift, grouped ordering assumptions, or open/unattended scope changes. | Focused `LiveReportHandler` and Analytics live metrics tests, full `go test ./...`, and `git diff --check` must pass. No route artifacts change. |
 | 0 | P3.47 live reports grouped error-shape parity | Implemented for reused live-report grouped metrics clients: GoChat now follows inspected Chatwoot `Api::V2::Accounts::LiveReportsController#set_group_scope` and `spec/controllers/api/v2/accounts/live_reports_controller_spec.rb` by returning raw `422 { error: "invalid group_by" }` for invalid grouped live-report requests. Focused tests also cover assignee grouping payloads, including the nil-assignee bucket, open/unattended/unassigned counts, and absence of the local success envelope. | Keep in Review; reopen from B12 live-report smoke or fresh reference evidence for authorization status drift, team lookup `404` behavior, grouped ordering assumptions, or pending/open scope drift. | Focused `LiveReportHandler` tests, full `go test ./...`, and `git diff --check` must pass. No route artifacts change. |
 | 0 | P3.46 report summary aggregate parity | Implemented for reused v2 reports summary clients: GoChat now follows inspected Chatwoot `Api::V2::Accounts::ReportsController#build_summary`, `V2::Reports::Conversations::MetricBuilder#summary/#bot_summary`, `Reports::RawDataSource#aggregate`, and `Reports::ReportMetricRegistry`. `/reports/summary` returns raw Chatwoot metric keys with a `previous` object instead of the local `{ metrics: [...] }` rollup envelope, uses current/previous Unix-second ranges, honors report `type/id` filters, and switches average values to `value_in_business_hours` when requested. `/reports/bot_summary` now applies aggregate raw count strategies: handoffs are distinct non-null conversations, and bot resolutions count event rows after excluding handoff conversations in the same scoped range. | Keep in Review; reopen from B12 reports summary smoke, fresh reference evidence for rollup-backed `Reports::DataSource.for`, invalid timezone/range controller behavior, or summary dimension scope drift for team/label/report-type filters. | Focused Analytics service bot-summary tests, Analytics handler summary tests, escalated full `go test ./...`, and `git diff --check` must pass. No route artifacts change. |
@@ -287,6 +288,7 @@ This ledger records the committed parity checkpoints that future slices should b
 
 | Commit | Scope | Verification summary | Follow-up state |
 | --- | --- | --- | --- |
+| `feat(reports): align conversation agent metrics` | Advances P3.49 reports conversation agent-metric parity by matching Chatwoot `Api::V2::Accounts::ReportsController#conversations` and `V2::ReportBuilder#conversation_metrics/#agent_metrics`. GoChat now returns empty `422` for blank `type`, keeps raw account live metrics for `type=account`, and returns paginated account-user agent objects for non-account requests instead of grouped assignee/team buckets; each agent row includes Chatwoot user fields plus `metric.open` and `metric.unattended`, sorted by open count descending. | `go test ./internal/handler/api/v1 -run AnalyticsHandler -count=1`; `go test ./internal/service -run AnalyticsReportsConversationsAgentMetrics -count=1`; full `go test ./...`; `git diff --check`. Route artifacts are unchanged. | P3.49 moves to Review for current reports conversation evidence; continue Phase 2/3 drift audit, Phase 6 placeholder audit, B12 live smoke, or fresh reference/smoke drift. |
 | `feat(reports): filter grouped live metrics by team` | Advances P3.48 live reports grouped team-scope parity by matching Chatwoot `Api::V2::Accounts::LiveReportsController#load_conversations`. GoChat grouped live-report metrics now parse optional `team_id` and apply that scope before grouping by `team_id` or `assignee_id`, so grouped assignee payloads exclude other-team conversations while retaining Chatwoot open/unattended/unassigned count fields and nil-assignee buckets. | `go test ./internal/handler/api/v1 -run LiveReportHandler -count=1`; `go test ./internal/service -run AnalyticsLiveConversationMetrics -count=1`; full `go test ./...`; `git diff --check`. Route artifacts are unchanged. | P3.48 moves to Review for current live-report team-scope evidence; continue Phase 2/3 drift audit, Phase 6 placeholder audit, B12 live smoke, or fresh reference/smoke drift. |
 | `feat(reports): align live grouped errors` | Advances P3.47 live reports grouped error-shape parity by matching Chatwoot `Api::V2::Accounts::LiveReportsController#set_group_scope` and live-report controller specs. GoChat now returns raw `422 { error: "invalid group_by" }` for invalid `/live_reports/grouped_conversation_metrics` requests instead of the local error envelope, and focused tests cover grouped assignee payloads with nil-assignee buckets and open/unattended/unassigned counts. | `go test ./internal/handler/api/v1 -run LiveReportHandler -count=1`; full `go test ./...`; `git diff --check`. Route artifacts are unchanged. | P3.47 moves to Review for current live-report grouped error evidence; continue Phase 2/3 drift audit, Phase 6 placeholder audit, B12 live smoke, or fresh reference/smoke drift. |
 | `feat(reports): align summary aggregates` | Advances P3.46 report summary aggregate parity by matching Chatwoot `Api::V2::Accounts::ReportsController#build_summary`, `V2::Reports::Conversations::MetricBuilder`, `Reports::RawDataSource#aggregate`, and `ReportMetricRegistry`. GoChat `/reports/summary` now returns raw current metric keys plus `previous`, computes current/previous range aggregates from persisted conversations/messages/reporting events, honors report `type/id` dimensions and business-hours averages, and makes `bot_summary` aggregate counts follow the same `distinct_conversation` and `exclude_bot_handoffs` raw count strategies as the current timeseries path. | `go test ./internal/service -run 'Analytics.*Summary\|BotSummary\|BotTimeseries\|AnalyticsReports' -count=1`; `go test ./internal/handler/api/v1 -run AnalyticsHandler -count=1`; escalated full `go test ./...`; `git diff --check`. Route artifacts are unchanged. | P3.46 moves to Review for current summary aggregate evidence; continue Phase 2/3 drift audit for report builder depth, Phase 6 placeholder audit, B12 live smoke, or fresh reference/smoke drift. |
@@ -2554,3 +2556,4 @@ Verification milestone gates:
 - 2026-06-07: P3.46 report summary aggregate checkpoint prepared as `feat(reports): align summary aggregates`; audited Chatwoot v2 reports controller summary builder, conversation metric builder, raw data-source aggregate, metric registry, and reused dashboard `reports.js` summary/bot-summary callers. GoChat `/reports/summary` now returns raw Chatwoot metric keys plus `previous`, derives current/previous aggregates from persisted rows, honors report `type/id` filters and business-hours averages, and fixes `/reports/bot_summary` aggregate counting so handoffs are distinct non-null conversations while bot resolutions count event rows after excluding handoff conversations. Focused Analytics service and handler tests plus escalated full `go test ./...` passed; `git diff --check` passed. No route artifacts change.
 - 2026-06-07: P3.47 live reports grouped error checkpoint prepared as `feat(reports): align live grouped errors`; audited Chatwoot v2 live reports controller `set_group_scope` and request specs. GoChat grouped live reports now return raw `422 { error: "invalid group_by" }` for invalid `group_by` instead of the local error envelope, and focused handler tests cover assignee grouping with nil-assignee buckets and Chatwoot open/unattended/unassigned counts. Focused `LiveReportHandler` tests passed; full `go test ./...` and `git diff --check` passed. No route artifacts change.
 - 2026-06-07: P3.48 live reports grouped team-scope checkpoint prepared as `feat(reports): filter grouped live metrics by team`; audited Chatwoot v2 live reports controller `load_conversations` and grouped metrics behavior. GoChat grouped live reports now parse optional `team_id` and apply that scope before grouping by assignee/team, so other-team conversations no longer leak into `grouped_conversation_metrics` when the frontend supplies a team filter. Focused `LiveReportHandler` and Analytics live metrics tests passed; full `go test ./...` and `git diff --check` passed. No route artifacts change.
+- 2026-06-07: P3.49 reports conversation agent-metric checkpoint prepared as `feat(reports): align conversation agent metrics`; audited Chatwoot v2 reports controller `conversations`, `V2::ReportBuilder#conversation_metrics/#agent_metrics`, `ReportHelper#scope`, and reused dashboard `reports.js` conversation metric caller. GoChat `/reports/conversations` now returns empty `422` for blank `type`, keeps raw account live metrics for `type=account`, and returns Chatwoot agent metric rows for non-account requests with paginated account users, availability, thumbnail, and `metric.open/unattended` sorted by open count instead of grouped assignee buckets. Focused Analytics handler and service tests passed; escalated full `go test ./...` and `git diff --check` passed. No route artifacts change.