feat(errors): per-account error modal with persisted error list

Accounts finishing done_with_errors showed only a count and a single
last_error. This adds a modal listing every concrete error of the
account's most recent run.

- migration 0005: account_errors table (kind folder|message|account,
  folder, message_ref, error, created_at; ON DELETE CASCADE; indexed)
- store: AddAccountError / ClearAccountErrors / ListAccountErrors
- copy: OnError callback captures per-message error text (previously
  only counted), with a "UID N: subject" reference
- orchestrator: clear errors at run start; persist folder/message/
  account errors; cap 500 rows/account/run with a suppressed-note row
- api: GET /api/tasks/{id}/accounts/{accountId}/errors
- web: AccountErrorsModal, clickable ERRORS count, api + styles

Verified: migration applies on Postgres 18; store add/list/clear and
cascade tests pass against real pg; backend build/vet/test green; web
tsc+vite build and oxlint clean.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01U9Eq4JtWjyNTv5qat3B3mM

internal/httpapi/accounts.go

@@ -226,6 +226,31 @@ func (s *Server) handleProbeAccountFolders(w http.ResponseWriter, r *http.Reques
 	})
 }
 
+// handleListAccountErrors returns the individual errors recorded for an account
+// during its most recent run, for the per-account error modal.
+func (s *Server) handleListAccountErrors(w http.ResponseWriter, r *http.Request) {
+	taskID, err := pathID(r, "id")
+	if err != nil {
+		http.Error(w, "bad id", http.StatusBadRequest)
+		return
+	}
+	accID, err := pathID(r, "accountId")
+	if err != nil {
+		http.Error(w, "bad account id", http.StatusBadRequest)
+		return
+	}
+	if _, ok := s.findAccount(r, taskID, accID); !ok {
+		http.Error(w, "account not found", http.StatusNotFound)
+		return
+	}
+	errs, err := s.store.ListAccountErrors(r.Context(), accID)
+	if err != nil {
+		http.Error(w, err.Error(), http.StatusInternalServerError)
+		return
+	}
+	writeJSON(w, http.StatusOK, errs)
+}
+
 // handleSetAccountFolderMapping persists one account's rename map + excluded set.
 func (s *Server) handleSetAccountFolderMapping(w http.ResponseWriter, r *http.Request) {
 	taskID, err := pathID(r, "id")

internal/httpapi/router.go

@@ -24,6 +24,7 @@ func (s *Server) Router() http.Handler {
 	api.HandleFunc("PUT /api/tasks/{id}/folder-mapping", s.handleSetFolderMapping)
 	api.HandleFunc("PUT /api/tasks/{id}/schedule", s.handleSetSchedule)
 	api.HandleFunc("GET /api/tasks/{id}/runs", s.handleListRuns)
+	api.HandleFunc("GET /api/tasks/{id}/accounts/{accountId}/errors", s.handleListAccountErrors)
 	api.HandleFunc("DELETE /api/tasks/{id}/accounts/{accountId}", s.handleDeleteAccount)
 	api.HandleFunc("POST /api/tasks/{id}/import", s.handleImportCSV)
 	api.HandleFunc("POST /api/tasks/{id}/test", s.handleTestAccounts)

internal/imapx/copy.go

@@ -27,6 +27,11 @@ type CopyDeps struct {
 	// folder's messages have been examined so far — so the UI shows movement
 	// while dedup decisions are made, before bodies start copying.
 	OnScan func(scanned, total int64)
+	// OnError is called for each message-level error, with a message reference
+	// ("UID N: subject", empty when the envelope is unavailable) and the error
+	// text — so the orchestrator can persist individual errors for the
+	// per-account error modal. Folder-level errors are reported by the caller.
+	OnError func(ref, msg string)
 }
 
 // CopyResult summarizes the outcome of one CopyFolder run.
@@ -44,6 +49,15 @@ type CopyResult struct {
 // checked between windows.
 const metaScanBatch = 1000
 
+// msgRef builds a human-readable reference for a message error: its UID plus
+// subject when known, e.g. "UID 42: Invoice". Falls back to just the UID.
+func msgRef(uid imap.UID, subject string) string {
+	if subject == "" {
+		return fmt.Sprintf("UID %d", uid)
+	}
+	return fmt.Sprintf("UID %d: %s", uid, subject)
+}
+
 // metaBatches tiles 1..total into contiguous, non-overlapping windows of at
 // most batchSize, covering every sequence number exactly once.
 func metaBatches(total, batchSize uint32) []imap.SeqRange {
@@ -95,11 +109,17 @@ func CopyFolder(ctx context.Context, src, dst *imapclient.Client, srcFolder, dst
 	type queued struct {
 		uid          imap.UID
 		key          string
+		subject      string
 		flags        []imap.Flag
 		internalDate time.Time
 	}
 	var todo []queued
 	var scanned int64
+	reportErr := func(ref, msg string) {
+		if deps.OnError != nil {
+			deps.OnError(ref, msg)
+		}
+	}
 	// Scan metadata in bounded windows instead of one FETCH 1:*, so each
 	// command is short (the server stays responsive) and ctx is checked on
 	// every window boundary — not just between messages of one giant command.
@@ -122,20 +142,26 @@ func CopyFolder(ctx context.Context, src, dst *imapclient.Client, srcFolder, dst
 			buf, err := msg.Collect()
 			if err != nil {
 				res.Errors++
+				reportErr("", "read message metadata: "+err.Error())
 				continue
 			}
 			scanned++
 			key := MessageKey(buf.Envelope, buf.RFC822Size)
+			subject := ""
+			if buf.Envelope != nil {
+				subject = buf.Envelope.Subject
+			}
 			already, err := deps.IsMigrated(key)
 			if err != nil {
 				res.Errors++
+				reportErr(msgRef(buf.UID, subject), "dedup lookup: "+err.Error())
 			} else if already {
 				res.Skipped++
 				if deps.OnProgress != nil {
 					deps.OnProgress(res.Copied, res.Skipped)
 				}
 			} else {
-				todo = append(todo, queued{uid: buf.UID, key: key, flags: buf.Flags, internalDate: buf.InternalDate})
+				todo = append(todo, queued{uid: buf.UID, key: key, subject: subject, flags: buf.Flags, internalDate: buf.InternalDate})
 			}
 			if deps.OnScan != nil {
 				deps.OnScan(scanned, total)
@@ -153,10 +179,12 @@ func CopyFolder(ctx context.Context, src, dst *imapclient.Client, srcFolder, dst
 		}
 		if err := streamOne(src, dst, dstFolder, q.uid, q.flags, q.internalDate); err != nil {
 			res.Errors++
+			reportErr(msgRef(q.uid, q.subject), "copy message: "+err.Error())
 			continue
 		}
 		if err := deps.MarkMigrated(dstFolder, q.key); err != nil {
 			res.Errors++
+			reportErr(msgRef(q.uid, q.subject), "mark migrated: "+err.Error())
 			continue
 		}
 		res.Copied++

internal/orchestrator/orchestrator.go

@@ -16,6 +16,12 @@ import (
 var ErrNotTested = errors.New("accounts not fully tested")
 var ErrAlreadyRunning = errors.New("task already running")
 
+// maxAccountErrors caps how many individual error rows one account records per
+// run, so a corrupt mailbox producing thousands of failures can't bloat the
+// account_errors table. The cap'th row is a synthetic "further errors
+// suppressed" note.
+const maxAccountErrors = 500
+
 // folderPlan is one source folder scheduled for copy and its destination name.
 type folderPlan struct {
 	src, dst string
@@ -264,6 +270,7 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 	_ = o.store.SetAccountStatus(ctx, a.ID, "running")
 	_ = o.store.SetAccountError(ctx, a.ID, "")     // clear any error from a previous run
 	_ = o.store.ResetAccountCounters(ctx, a.ID)    // start from zero; IncAccountCounters is additive
+	_ = o.store.ClearAccountErrors(ctx, a.ID)      // drop last run's per-error rows
 
 	// Per-account cancellable context: IMAP work uses actx (so CancelAccount
 	// stops it); DB writes keep the parent ctx so status/counters persist even
@@ -277,28 +284,28 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 
 	srcPass, err := crypto.Decrypt(o.encKey, a.SrcPassEnc)
 	if err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "src", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "src", err)
 	}
 	dstPass, err := crypto.Decrypt(o.encKey, a.DstPassEnc)
 	if err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "dst", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "dst", err)
 	}
 
 	src, err := imapx.Connect(actx, srcEP)
 	if err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "src", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "src", err)
 	}
 	defer func() { _ = src.Logout().Wait() }()
 	if err := src.Login(a.SrcLogin, string(srcPass)).Wait(); err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "src", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "src", err)
 	}
 	dst, err := imapx.Connect(actx, dstEP)
 	if err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "dst", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "dst", err)
 	}
 	defer func() { _ = dst.Logout().Wait() }()
 	if err := dst.Login(a.DstLogin, string(dstPass)).Wait(); err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "dst", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "dst", err)
 	}
 
 	// On cancel, close the connections so any in-flight network read (a slow
@@ -311,7 +318,7 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 
 	folders, err := imapx.ListFolders(src)
 	if err != nil {
-		return o.accountFailed(ctx, task.ID, a, srcEP, dstEP, "src", err)
+		return o.accountFailed(ctx, task.ID, runID, a, srcEP, dstEP, "src", err)
 	}
 
 	// Planning pass: decide folders from the account's own config, then EXAMINE
@@ -341,9 +348,24 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 	var curFolder string
 	var curTotal int64
 	var lastEmit, lastScanEmit time.Time
+	// Persist individual errors for the per-account error modal, capped so a
+	// corrupt mailbox can't write unbounded rows. Runs on this goroutine, so
+	// the counter is race-free (same reasoning as the progress vars above).
+	var persistedErrs int
+	addErr := func(kind, folder, ref, msg string) {
+		persistedErrs++
+		switch {
+		case persistedErrs < maxAccountErrors:
+			_ = o.store.AddAccountError(ctx, a.ID, runID, kind, folder, ref, msg)
+		case persistedErrs == maxAccountErrors:
+			_ = o.store.AddAccountError(ctx, a.ID, runID, "account", "", "",
+				"too many errors — further errors suppressed")
+		}
+	}
 	deps := imapx.CopyDeps{
 		IsMigrated:   func(k string) (bool, error) { return o.store.IsMigrated(ctx, a.ID, k) },
 		MarkMigrated: func(folder, k string) error { return o.store.MarkMigrated(ctx, a.ID, folder, k) },
+		OnError:      func(ref, msg string) { addErr("message", curFolder, ref, msg) },
 		OnProgress: func(c, s int) {
 			now := time.Now()
 			done := c + s
@@ -393,6 +415,7 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 			slog.Warn("folder copy error", "account", a.ID, "src_login", a.SrcLogin, "folder", fp.src, "err", err)
 			folderErr = 1
 			_ = o.store.SetAccountError(ctx, a.ID, "folder \""+fp.src+"\": "+err.Error())
+			addErr("folder", fp.src, "", err.Error())
 			o.hub.Publish(wshub.Event{Type: "error", TaskID: task.ID, Data: map[string]any{
 				"account_id": a.ID, "src_login": a.SrcLogin, "folder": fp.src, "error": err.Error(),
 			}})
@@ -428,7 +451,7 @@ func (o *Orchestrator) runAccount(ctx context.Context, task store.Task, runID in
 	return copied, skipped, errs
 }
 
-func (o *Orchestrator) accountFailed(ctx context.Context, taskID int64, a store.Account, srcEP, dstEP imapx.Endpoint, side string, err error) (int64, int64, int64) {
+func (o *Orchestrator) accountFailed(ctx context.Context, taskID, runID int64, a store.Account, srcEP, dstEP imapx.Endpoint, side string, err error) (int64, int64, int64) {
 	// A cancellation surfacing as an error is a cancel, not a failure.
 	if errors.Is(err, context.Canceled) {
 		_ = o.store.SetAccountStatus(ctx, a.ID, "cancelled")
@@ -442,7 +465,9 @@ func (o *Orchestrator) accountFailed(ctx context.Context, taskID int64, a store.
 	}
 	slog.Error("account failed", "account", a.ID, "side", side, "login", login, "host", host, "port", port, "err", err)
 	_ = o.store.SetAccountStatus(ctx, a.ID, "error")
-	_ = o.store.SetAccountError(ctx, a.ID, side+" "+login+"@"+host+": "+err.Error())
+	failMsg := side + " " + login + "@" + host + ": " + err.Error()
+	_ = o.store.SetAccountError(ctx, a.ID, failMsg)
+	_ = o.store.AddAccountError(ctx, a.ID, runID, "account", "", "", failMsg)
 	o.hub.Publish(wshub.Event{Type: "error", TaskID: taskID,
 		Data: map[string]any{"account_id": a.ID, "side": side, "login": login, "host": host, "port": port, "error": err.Error()}})
 	return 0, 0, 1

internal/store/account_errors.go

@@ -0,0 +1,61 @@
+package store
+
+import (
+	"context"
+	"time"
+)
+
+// AccountError is one concrete error recorded during an account's run —
+// folder-level, message-level, or account-level (connect/login).
+type AccountError struct {
+	ID         int64     `json:"id"`
+	AccountID  int64     `json:"account_id"`
+	RunID      int64     `json:"run_id"`
+	Kind       string    `json:"kind"` // folder | message | account
+	Folder     string    `json:"folder"`
+	MessageRef string    `json:"message_ref"`
+	Error      string    `json:"error"`
+	CreatedAt  time.Time `json:"created_at"`
+}
+
+// AddAccountError appends one error row for an account's current run.
+func (s *Store) AddAccountError(ctx context.Context, accountID, runID int64, kind, folder, ref, msg string) error {
+	_, err := s.Pool.Exec(ctx,
+		`INSERT INTO account_errors (account_id, run_id, kind, folder, message_ref, error)
+		 VALUES ($1,$2,$3,$4,$5,$6)`,
+		accountID, runID, kind, folder, ref, msg)
+	return err
+}
+
+// ClearAccountErrors removes an account's errors at the start of a run, so the
+// list reflects only the current run (mirrors ResetAccountCounters).
+func (s *Store) ClearAccountErrors(ctx context.Context, accountID int64) error {
+	_, err := s.Pool.Exec(ctx, `DELETE FROM account_errors WHERE account_id=$1`, accountID)
+	return err
+}
+
+// ListAccountErrors returns an account's errors in insertion order, for the
+// per-account error modal.
+func (s *Store) ListAccountErrors(ctx context.Context, accountID int64) ([]AccountError, error) {
+	rows, err := s.Pool.Query(ctx,
+		`SELECT id, account_id, run_id, kind, folder, message_ref, error, created_at
+		 FROM account_errors WHERE account_id=$1 ORDER BY id`, accountID)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+	out := []AccountError{}
+	for rows.Next() {
+		var e AccountError
+		var runID *int64
+		if err := rows.Scan(&e.ID, &e.AccountID, &runID, &e.Kind, &e.Folder,
+			&e.MessageRef, &e.Error, &e.CreatedAt); err != nil {
+			return nil, err
+		}
+		if runID != nil {
+			e.RunID = *runID
+		}
+		out = append(out, e)
+	}
+	return out, rows.Err()
+}

internal/store/account_errors_test.go

@@ -0,0 +1,75 @@
+package store
+
+import (
+	"context"
+	"testing"
+)
+
+// AddAccountError persists individual errors; ListAccountErrors returns them in
+// insertion order; ClearAccountErrors wipes them for the next run.
+func TestAccountErrorsAddListClear(t *testing.T) {
+	s := testStore(t)
+	ctx := context.Background()
+	epSrc, _ := s.CreateEndpoint(ctx, Endpoint{RoleLabel: "src", Host: "a", Port: 993, TLSMode: "ssl"})
+	epDst, _ := s.CreateEndpoint(ctx, Endpoint{RoleLabel: "dst", Host: "b", Port: 993, TLSMode: "ssl"})
+	taskID, _ := s.CreateTask(ctx, Task{Name: "t", SrcEndpointID: epSrc, DstEndpointID: epDst})
+	runID, _ := s.CreateRun(ctx, taskID, "manual")
+	accID, _ := s.CreateAccount(ctx, Account{TaskID: taskID, SrcLogin: "u", SrcPassEnc: "x", DstLogin: "u2", DstPassEnc: "y"})
+
+	if err := s.AddAccountError(ctx, accID, runID, "folder", "INBOX", "", "examine failed"); err != nil {
+		t.Fatalf("add folder: %v", err)
+	}
+	if err := s.AddAccountError(ctx, accID, runID, "message", "INBOX", "UID 42: hi", "append rejected"); err != nil {
+		t.Fatalf("add message: %v", err)
+	}
+
+	errs, err := s.ListAccountErrors(ctx, accID)
+	if err != nil {
+		t.Fatalf("list: %v", err)
+	}
+	if len(errs) != 2 {
+		t.Fatalf("len=%d want 2", len(errs))
+	}
+	if errs[0].Kind != "folder" || errs[0].Folder != "INBOX" || errs[0].Error != "examine failed" {
+		t.Fatalf("errs[0]=%+v", errs[0])
+	}
+	if errs[1].Kind != "message" || errs[1].MessageRef != "UID 42: hi" || errs[1].Error != "append rejected" {
+		t.Fatalf("errs[1]=%+v", errs[1])
+	}
+	if errs[0].ID >= errs[1].ID {
+		t.Fatalf("expected insertion order by id: %d then %d", errs[0].ID, errs[1].ID)
+	}
+
+	if err := s.ClearAccountErrors(ctx, accID); err != nil {
+		t.Fatalf("clear: %v", err)
+	}
+	errs, err = s.ListAccountErrors(ctx, accID)
+	if err != nil {
+		t.Fatalf("list after clear: %v", err)
+	}
+	if len(errs) != 0 {
+		t.Fatalf("after clear len=%d want 0", len(errs))
+	}
+}
+
+// Deleting an account cascades its errors (ON DELETE CASCADE).
+func TestAccountErrorsCascadeOnAccountDelete(t *testing.T) {
+	s := testStore(t)
+	ctx := context.Background()
+	epSrc, _ := s.CreateEndpoint(ctx, Endpoint{RoleLabel: "src", Host: "a", Port: 993, TLSMode: "ssl"})
+	epDst, _ := s.CreateEndpoint(ctx, Endpoint{RoleLabel: "dst", Host: "b", Port: 993, TLSMode: "ssl"})
+	taskID, _ := s.CreateTask(ctx, Task{Name: "t", SrcEndpointID: epSrc, DstEndpointID: epDst})
+	accID, _ := s.CreateAccount(ctx, Account{TaskID: taskID, SrcLogin: "u", SrcPassEnc: "x", DstLogin: "u2", DstPassEnc: "y"})
+	_ = s.AddAccountError(ctx, accID, 0, "account", "", "", "login failed")
+
+	if err := s.DeleteAccount(ctx, accID); err != nil {
+		t.Fatalf("delete: %v", err)
+	}
+	var n int
+	if err := s.Pool.QueryRow(ctx, `SELECT count(*) FROM account_errors WHERE account_id=$1`, accID).Scan(&n); err != nil {
+		t.Fatalf("count: %v", err)
+	}
+	if n != 0 {
+		t.Fatalf("account_errors not cascaded: %d rows", n)
+	}
+}

internal/store/store_test.go

@@ -17,7 +17,7 @@ func testStore(t *testing.T) *Store {
 	}
 	t.Cleanup(func() {
 		s.Pool.Exec(context.Background(),
-			`TRUNCATE endpoints, tasks, accounts, runs, migrated_messages RESTART IDENTITY CASCADE`)
+			`TRUNCATE endpoints, tasks, accounts, runs, migrated_messages, account_errors RESTART IDENTITY CASCADE`)
 		s.Pool.Close()
 	})
 	return s