diff --git a/assets/tui.gif b/assets/tui.gif
index 216d546..1e7f0cf 100644
Binary files a/assets/tui.gif and b/assets/tui.gif differ
diff --git a/assets/vhs/tui.tape b/assets/vhs/tui.tape
index f3257d3..64091fd 100644
--- a/assets/vhs/tui.tape
+++ b/assets/vhs/tui.tape
@@ -34,5 +34,9 @@ Sleep 1s
 Enter
 Sleep 1.5s
 
+# scroll down
+Down 10
+Sleep 0.8s
+
 Type "q"
 Sleep 750ms
diff --git a/internal/cli/render/render.go b/internal/cli/render/render.go
index 14f7d07..e772dad 100644
--- a/internal/cli/render/render.go
+++ b/internal/cli/render/render.go
@@ -284,8 +284,12 @@ func SummaryHuman(w io.Writer, st Style, s core.Summary) error {
 		fmt.Fprintf(w, "\n%s\n", st.Bold(fmt.Sprintf("Open audits (%d)", len(s.OpenAudits))))
 		rows := make([][]string, 0, len(s.OpenAudits))
 		for _, a := range s.OpenAudits {
-			bar := fmt.Sprintf("%s %s", st.SegmentBar(a.DoneFindings, a.ActiveFindings, a.DroppedFindings, a.Findings, 10), st.Percent(a.Percent()))
-			rows = append(rows, []string{"  " + st.Bold(a.Slug), bar, theme.Counts(a.Resolved(), a.Findings), a.Area})
+			bar := fmt.Sprintf("%s %s", st.SegmentBar(a.DoneFindings, a.ActiveFindings, a.DroppedFindings, a.Findings, 10), st.AuditPercent(a.Percent()))
+			counts := theme.Counts(a.Resolved(), a.Findings)
+			if note := auditStateNote(st, a, false); note != "" {
+				counts += "  " + note
+			}
+			rows = append(rows, []string{"  " + st.Bold(a.Slug), bar, counts, a.Area})
 		}
 		writeTable(w, st.width, nil, rows)
 	}
@@ -560,6 +564,24 @@ func EpicShowHuman(w io.Writer, st Style, es core.EpicSummary, tasks []domain.Ta
 	return nil
 }
 
+// auditStateNote is the trailing call-to-action on an audit's progress line, kept
+// in one place so the list / show / dashboard surfaces can't drift. An open,
+// fully-triaged audit is flagged "✔ ready to close" (green) on every surface — the
+// distinguisher that keeps a settled audit (full bar, 0% fixed) from reading like an
+// untouched one. detail=true (single-audit views) additionally shows the routine
+// "(N open)" while findings remain; list rows omit that to stay scannable. Returns
+// "" (no leading space) when there is nothing to note.
+func auditStateNote(st Style, a domain.Audit, detail bool) string {
+	switch {
+	case a.ReadyToClose():
+		return st.Green("✔ ready to close")
+	case detail && a.OpenFindings > 0:
+		return fmt.Sprintf("(%d open)", a.OpenFindings)
+	default:
+		return ""
+	}
+}
+
 // AuditsHuman writes a table of audits with finding counts.
 func AuditsHuman(w io.Writer, st Style, audits []domain.Audit) error {
 	if len(audits) == 0 {
@@ -568,7 +590,10 @@ func AuditsHuman(w io.Writer, st Style, audits []domain.Audit) error {
 	rows := make([][]string, 0, len(audits))
 	for _, a := range audits {
 		bar := st.SegmentBar(a.DoneFindings, a.ActiveFindings, a.DroppedFindings, a.Findings, 8)
-		progress := fmt.Sprintf("%s %s %s", bar, st.Percent(a.Percent()), theme.Counts(a.Resolved(), a.Findings))
+		progress := fmt.Sprintf("%s %s %s", bar, st.AuditPercent(a.Percent()), theme.Counts(a.Resolved(), a.Findings))
+		if note := auditStateNote(st, a, false); note != "" {
+			progress += "  " + note
+		}
 		rows = append(rows, []string{st.Bucket(string(a.Bucket)), st.Bold(a.Slug), progress, a.Area})
 	}
 	writeTable(w, st.width, []string{st.Dim("BUCKET"), st.Dim("AUDIT"), st.Dim("PROGRESS"), st.Dim("AREA")}, rows)
@@ -610,9 +635,9 @@ func AuditShowHuman(w io.Writer, st Style, a domain.Audit, findings []domain.Fin
 		field("updated", a.Updated)
 	}
 	bar := st.SegmentBar(a.DoneFindings, a.ActiveFindings, a.DroppedFindings, a.Findings, 10)
-	progress := fmt.Sprintf("%s %s  %s", bar, st.Percent(a.Percent()), theme.Counts(a.Resolved(), a.Findings))
-	if a.OpenFindings > 0 {
-		progress += fmt.Sprintf("  (%d open)", a.OpenFindings)
+	progress := fmt.Sprintf("%s %s  %s", bar, st.AuditPercent(a.Percent()), theme.Counts(a.Resolved(), a.Findings))
+	if note := auditStateNote(st, a, true); note != "" {
+		progress += "  " + note
 	}
 	field("findings", progress)
 	// A status-grouped finding tree (glyph + code + title), mirroring epic show's
diff --git a/internal/cli/render/render_test.go b/internal/cli/render/render_test.go
index a4d3a50..64f2462 100644
--- a/internal/cli/render/render_test.go
+++ b/internal/cli/render/render_test.go
@@ -250,6 +250,56 @@ func TestAuditsJSONAndHuman(t *testing.T) {
 	}
 }
 
+// TestAuditProgressDistinguishesTriaged: the headline number must tell one story
+// with the bar. A fully-triaged open audit (all superseded, 0 fixed) reads
+// "0% fixed" like an untouched one, so the "ready to close" marker is what sets
+// them apart; an untouched audit shows neither the marker nor a fixed share.
+func TestAuditProgressDistinguishesTriaged(t *testing.T) {
+	triaged := domain.Audit{Slug: "2026-06-01-triaged", Bucket: domain.AuditOpen, Area: "store", Findings: 2, DroppedFindings: 2}
+	untouched := domain.Audit{Slug: "2026-06-01-untouched", Bucket: domain.AuditOpen, Area: "store", Findings: 2, OpenFindings: 2}
+
+	var out bytes.Buffer
+	if err := AuditsHuman(&out, NewStyle(false), []domain.Audit{triaged, untouched}); err != nil {
+		t.Fatal(err)
+	}
+	lines := strings.Split(strings.TrimRight(out.String(), "\n"), "\n")
+	var triagedLine, untouchedLine string
+	for _, ln := range lines {
+		switch {
+		case strings.Contains(ln, "triaged"):
+			triagedLine = ln
+		case strings.Contains(ln, "untouched"):
+			untouchedLine = ln
+		}
+	}
+	if !strings.Contains(triagedLine, "0% fixed") {
+		t.Errorf("triaged audit should label its share %q:\n%s", "0% fixed", triagedLine)
+	}
+	if !strings.Contains(triagedLine, "ready to close") {
+		t.Errorf("fully-triaged open audit should be flagged ready to close:\n%s", triagedLine)
+	}
+	if strings.Contains(untouchedLine, "ready to close") {
+		t.Errorf("untouched audit must NOT be flagged ready to close:\n%s", untouchedLine)
+	}
+
+	// audit show: an open audit with open findings keeps the "(N open)" note; a
+	// settled one gets the ready-to-close marker instead.
+	out.Reset()
+	if err := AuditShowHuman(&out, NewStyle(false), untouched, nil, ""); err != nil {
+		t.Fatal(err)
+	}
+	if s := out.String(); !strings.Contains(s, "(2 open)") || strings.Contains(s, "ready to close") {
+		t.Errorf("audit show for an untouched audit should note (2 open), not ready to close:\n%s", s)
+	}
+	out.Reset()
+	if err := AuditShowHuman(&out, NewStyle(false), triaged, nil, ""); err != nil {
+		t.Fatal(err)
+	}
+	if s := out.String(); !strings.Contains(s, "ready to close") || strings.Contains(s, "open)") {
+		t.Errorf("audit show for a triaged audit should be ready to close, no open note:\n%s", s)
+	}
+}
+
 // TestAuditShowHuman_FindingTree: audit show renders meta + a status-grouped
 // finding tree (lifecycle order, glyph-coded) + the body, mirroring epic show.
 // A finding with no status must land in a trailing group, not vanish.
diff --git a/internal/cli/render/style.go b/internal/cli/render/style.go
index 02cbd9f..8c42d2a 100644
--- a/internal/cli/render/style.go
+++ b/internal/cli/render/style.go
@@ -170,6 +170,13 @@ func (s Style) Percent(pct int) string {
 	return s.wrap(s.colorSeq(theme.Percent(pct)), theme.PercentLabel(pct))
 }
 
+// AuditPercent colors an audit's fixed-share percent on the same gray/yellow/green
+// tiers as Percent, but labels it "N% fixed" (theme.AuditPercentLabel) so the
+// number reads as the fixed share, not overall triage.
+func (s Style) AuditPercent(pct int) string {
+	return s.wrap(s.colorSeq(theme.Percent(pct)), theme.AuditPercentLabel(pct))
+}
+
 // Green / Red / Warn style status glyphs and success/error/warning text.
 func (s Style) Green(t string) string { return s.wrap(s.colorSeq(theme.ColorGreen), t) }
 func (s Style) Red(t string) string   { return s.wrap(s.colorSeq(theme.ColorRed), t) }
diff --git a/internal/domain/audit.go b/internal/domain/audit.go
index 1b2a4c9..0d7cfb8 100644
--- a/internal/domain/audit.go
+++ b/internal/domain/audit.go
@@ -100,3 +100,9 @@ func (a Audit) Percent() int {
 func (a Audit) Settled() bool {
 	return a.Findings > 0 && a.DoneFindings+a.DroppedFindings == a.Findings
 }
+
+// ReadyToClose is the call-to-action shared by the --json envelope (ready_to_close)
+// and the human progress line: an OPEN audit that is Settled has no findings left
+// to work and can be closed. A closed/deferred audit is not "ready to close" (it is
+// already off the open board), so this is false there regardless of Settled.
+func (a Audit) ReadyToClose() bool { return a.Bucket == AuditOpen && a.Settled() }
diff --git a/internal/domain/audit_settled_test.go b/internal/domain/audit_settled_test.go
index a05a88b..19789cd 100644
--- a/internal/domain/audit_settled_test.go
+++ b/internal/domain/audit_settled_test.go
@@ -21,3 +21,22 @@ func TestAuditSettled(t *testing.T) {
 		}
 	}
 }
+
+func TestAuditReadyToClose(t *testing.T) {
+	for _, c := range []struct {
+		name string
+		a    Audit
+		want bool
+	}{
+		{"open + all done", Audit{Bucket: AuditOpen, Findings: 3, DoneFindings: 3}, true},
+		{"open + all dropped (triaged)", Audit{Bucket: AuditOpen, Findings: 2, DroppedFindings: 2}, true},
+		{"open + still has open", Audit{Bucket: AuditOpen, Findings: 2, OpenFindings: 1, DoneFindings: 1}, false},
+		{"open + no findings", Audit{Bucket: AuditOpen, Findings: 0}, false},
+		{"closed but settled (already off the board)", Audit{Bucket: AuditClosed, Findings: 1, DoneFindings: 1}, false},
+		{"deferred but settled", Audit{Bucket: AuditDeferred, Findings: 2, DroppedFindings: 2}, false},
+	} {
+		if got := c.a.ReadyToClose(); got != c.want {
+			t.Errorf("%s: ReadyToClose() = %v, want %v", c.name, got, c.want)
+		}
+	}
+}
diff --git a/internal/theme/progress.go b/internal/theme/progress.go
index 4353450..7c4d9ef 100644
--- a/internal/theme/progress.go
+++ b/internal/theme/progress.go
@@ -17,6 +17,16 @@ func PercentLabel(pct int) string { return fmt.Sprintf("%d%%", pct) }
 // "100%") so it aligns in hand-laid-out columns / rows.
 func PercentLabelPadded(pct int) string { return fmt.Sprintf("%3d%%", pct) }
 
+// AuditPercentLabel qualifies an audit's percent as the FIXED share — "0% fixed"
+// — so a bare number can't be misread as overall progress. An epic's percent is
+// unambiguous (done/total), but an audit bands four dispositions: a fully-triaged
+// audit is legitimately 0% fixed yet ready to close, so its number names its unit.
+func AuditPercentLabel(pct int) string { return fmt.Sprintf("%d%% fixed", pct) }
+
+// AuditPercentLabelPadded is AuditPercentLabel right-justified to 3 percent digits
+// ("  0% fixed", "100% fixed") so audit list rows align.
+func AuditPercentLabelPadded(pct int) string { return fmt.Sprintf("%3d%% fixed", pct) }
+
 // Counts renders a done/total rollup ("7/12"). Width-justification for aligned
 // columns is the caller's concern (CLI tables pad cells; the TUI measures + pads).
 func Counts(done, total int) string { return fmt.Sprintf("%d/%d", done, total) }
diff --git a/internal/tui/detail.go b/internal/tui/detail.go
index 8ec557e..c68d07c 100644
--- a/internal/tui/detail.go
+++ b/internal/tui/detail.go
@@ -477,8 +477,11 @@ func renderAuditMeta(a domain.Audit, body string, width int, s *styles) string {
 	pct := a.Percent()
 	progress := fmt.Sprintf("%s %s  %s",
 		s.segBar(a.DoneFindings, a.ActiveFindings, a.DroppedFindings, a.Findings, 12),
-		s.fg(theme.Percent(pct), theme.PercentLabel(pct)), theme.Counts(a.Resolved(), a.Findings))
-	if a.OpenFindings > 0 {
+		s.fg(theme.Percent(pct), theme.AuditPercentLabel(pct)), theme.Counts(a.Resolved(), a.Findings))
+	switch {
+	case a.ReadyToClose():
+		progress += "  " + s.fg(theme.ColorGreen, "✔ ready to close")
+	case a.OpenFindings > 0:
 		progress += fmt.Sprintf("  (%d open)", a.OpenFindings)
 	}
 	detailField(&b, "audit", a.Slug, s)
diff --git a/internal/tui/item.go b/internal/tui/item.go
index 95ce249..d9bece7 100644
--- a/internal/tui/item.go
+++ b/internal/tui/item.go
@@ -214,8 +214,12 @@ func (d auditDelegate) Render(w io.Writer, m list.Model, index int, item list.It
 	tok := theme.Bucket(it.a.Bucket)
 	pct := it.a.Percent()
 	bar := st.segBar(it.a.DoneFindings, it.a.ActiveFindings, it.a.DroppedFindings, it.a.Findings, 8)
-	pctStr := st.fg(theme.Percent(pct), theme.PercentLabelPadded(pct))
+	pctStr := st.fg(theme.Percent(pct), theme.AuditPercentLabelPadded(pct))
 	counts := rollupCounts(it.a.Resolved(), it.a.Findings, it.countsW)
-	row(w, m, index, fmt.Sprintf("%s %s %s %s  %s  %s",
-		st.fg(tok.Color, tok.Glyph), bar, pctStr, counts, it.a.Slug, st.dim(it.a.Area)), st)
+	line := fmt.Sprintf("%s %s %s %s  %s  %s",
+		st.fg(tok.Color, tok.Glyph), bar, pctStr, counts, it.a.Slug, st.dim(it.a.Area))
+	if it.a.ReadyToClose() {
+		line += "  " + st.fg(theme.ColorGreen, "✔ ready to close")
+	}
+	row(w, m, index, line, st)
 }
diff --git a/internal/wire/dto.go b/internal/wire/dto.go
index 0efc395..187a8f0 100644
--- a/internal/wire/dto.go
+++ b/internal/wire/dto.go
@@ -183,7 +183,7 @@ func ToAuditJSON(a domain.Audit) AuditJSON {
 		ID: a.ID, Slug: a.Slug, Bucket: string(a.Bucket), Area: a.Area, Date: a.Date, Updated: a.Updated,
 		Findings: a.Findings, OpenFindings: a.OpenFindings,
 		InProgressFindings: a.ActiveFindings, DoneFindings: a.DoneFindings, DroppedFindings: a.DroppedFindings,
-		ReadyToClose: a.Bucket == domain.AuditOpen && a.Settled(),
+		ReadyToClose: a.ReadyToClose(),
 	}
 }
 
diff --git a/planning/tasks/6fq9zy102rea-audit-progress-number-contradicts-its-own-bar.md b/planning/tasks/6fq9zy102rea-audit-progress-number-contradicts-its-own-bar.md
new file mode 100644
index 0000000..6daa89a
--- /dev/null
+++ b/planning/tasks/6fq9zy102rea-audit-progress-number-contradicts-its-own-bar.md
@@ -0,0 +1,54 @@
+---
+schema: 1
+id: 6fq9zy102rea
+status: completed
+epic: 20-cli-ux-and-ergonomics
+description: audit show/list % counts only fixed+landed while the bar bands 4 states, so a fully-superseded (triaged) audit reads 0%.
+effort: Unknown
+tier: 3
+priority: high
+autonomy_level: 3
+tags: [audit, ux]
+created: "2026-07-18"
+updated_at: "2026-07-21"
+started_at: "2026-07-21"
+completed_at: "2026-07-21"
+---
+# Audit progress number contradicts its own bar
+
+## Objective
+
+`audit show` / `audit list` show a headline `N%` that counts only
+fixed + landed (`FindingTally.Done`, internal/domain/finding.go:146), but the
+segmented bar bands by four dispositions (open / in-progress / done / dropped).
+So the number and the bar disagree in one widget: an audit whose findings are all
+`superseded` (fully triaged into tasks, correctly closeable) shows `0%` —
+indistinguishable *at the number* from an untouched audit. Confirmed: two
+superseded findings render `▒▒▒▒▒▒▒▒▒▒ 0% 0/2`.
+
+## Acceptance criteria
+
+- [ ] The headline number and the bar tell one consistent story
+- [ ] A fully-triaged audit (0 open, all superseded) is distinguishable from an untouched one
+- [ ] The `audit list` `open` column reconciles with that same notion of progress
+
+## Notes
+
+- Options from the report: label it (`0% resolved`), show a stacked count
+  (`2 superseded · 2 in-progress · 0 fixed`), or make the number track the bar's gradient.
+- Loci: `TallyFindings`/`FindingTally` (internal/domain/finding.go:143), the
+  headline `%` (`Percent`/`SegmentBar` in internal/cli/render/style.go).
+- Source: https://github.com/andy-esch/taskflow/issues/105 (P1, High)
+
+## Resolution (2026-07-21)
+
+Kept the headline % meaning **fixed share** (never over-claims) and surfaced the
+already-computed settled state as a green `✔ ready to close` marker — the
+distinguisher between a triaged audit (full bar, 0% fixed) and an untouched one.
+
+- `domain.Audit.ReadyToClose()` centralizes the open+Settled call-to-action (deduped from the JSON DTO); JSON `ready_to_close` unchanged.
+- `theme.AuditPercentLabel`/`…Padded` label the audit number as `N% fixed` (epics keep the bare %).
+- One `auditStateNote` helper drives `audit show`/`list`/dashboard; TUI list row + detail mirror it, so no surface drifts.
+- `audit show`: `(N open)` while work remains, else `✔ ready to close`; list rows show only the ready marker to stay scannable.
+
+All three acceptance criteria met: number+bar tell one story; triaged vs untouched distinguishable; `open` column reconciles (0 open ⟺ ready to close). Tests + lint green.
diff --git a/planning/tasks/6fq9zy126tas-invalid-non-crockford-entity-id-is-mishandled.md b/planning/tasks/6fq9zy126tas-invalid-non-crockford-entity-id-is-mishandled.md
new file mode 100644
index 0000000..cbd2516
--- /dev/null
+++ b/planning/tasks/6fq9zy126tas-invalid-non-crockford-entity-id-is-mishandled.md
@@ -0,0 +1,38 @@
+---
+schema: 1
+id: 6fq9zy126tas
+status: ready-to-start
+epic: 24-data-model-evolution-stable-key-storage-read-model-content-occ
+description: A non-Crockford id is misreported as 'no leading id', aborts the whole command (exit 11), and lint --fix can't repair it.
+effort: Unknown
+tier: 3
+priority: high
+autonomy_level: 3
+tags: [id, store]
+created: "2026-07-18"
+---
+# Invalid (non-Crockford) entity id is mishandled
+
+## Objective
+
+An entity file whose leading id contains a non-Crockford char (Crockford base32
+excludes i/l/o/u) is misdiagnosed and unrepairable. `splitFlatName` calls
+`id.Valid` (internal/store/flatname.go:32); a bad char fails it, so the file is
+reported as *not id-led* even though a 12-char id is right there. Reproduced on
+current code with an `l` in the id.
+
+## Acceptance criteria
+
+- [ ] Error names the offending char / rule (e.g. `id "…" contains non-Crockford char 'l'`) instead of "has no leading id"
+- [ ] One bad file no longer forces exit 11 for the whole command — skip-with-warning (or gate the abort behind `--strict`) so unrelated queries succeed with a trustworthy exit code
+- [ ] `lint --fix` repairs/re-mints an invalid id, not only a missing one
+- [ ] Ids are validated at write time so a bad id can't persist
+
+## Notes
+
+- Today: `audit list` prints good rows then `error: … 1 file(s) with unreadable
+  frontmatter` (exit 11); `lint --fix` "fixes 1 file" but leaves the bad-id file untouched.
+- Manual fix that worked: swap the illegal char for its Crockford decode-alias
+  (l→1, i→1, o→0) in both filename and `id:` — same decoded value, canonical, keeps sort order.
+- Loci: internal/store/flatname.go, internal/id/id.go (Valid/decode), scanDir error surfacing.
+- Source: https://github.com/andy-esch/taskflow/issues/105 (P2, High)
diff --git a/planning/tasks/6fq9zy13wkdc-audit-lint-fix-for-legacy-finding-status-debt.md b/planning/tasks/6fq9zy13wkdc-audit-lint-fix-for-legacy-finding-status-debt.md
new file mode 100644
index 0000000..13f1402
--- /dev/null
+++ b/planning/tasks/6fq9zy13wkdc-audit-lint-fix-for-legacy-finding-status-debt.md
@@ -0,0 +1,35 @@
+---
+schema: 1
+id: 6fq9zy13wkdc
+status: ready-to-start
+epic: 20-cli-ux-and-ergonomics
+description: No audit lint --fix to normalize legacy finding statuses (emoji/legacy words); schema audit omits the status vocabulary.
+effort: Unknown
+tier: 3
+priority: medium
+autonomy_level: 3
+tags: [audit, lint]
+created: "2026-07-18"
+---
+# audit lint --fix for legacy finding-status debt + document the vocabulary
+
+## Objective
+
+`audit lint` enforces a strict finding-status vocabulary
+(deferred/fixed/in-progress/landed/open/superseded/wontfix) but there is no
+`audit lint --fix` to normalize older audits (emoji ✅/⏳/⛔, legacy words like
+`tracked`/`declined`, or pre-`Status:` findings). Separately, `schema audit`
+documents the Status-line *format* but not the allowed *vocabulary* (only the
+top-level `schema` lists it).
+
+## Acceptance criteria
+
+- [ ] `audit lint --fix` normalizes finding statuses (strip emoji, map declined→wontfix / tracked→superseded, backfill missing)
+- [ ] `schema audit` lists the finding-status vocabulary
+
+## Notes
+
+- Was masked in the wild by the P2 abort — whole-tree `audit lint` never
+  completed until the invalid-id files were fixed, hiding ~10 audits of debt.
+- Confirmed: no `--fix` on `audit lint`; `schema audit` omits the vocab.
+- Source: https://github.com/andy-esch/taskflow/issues/105 (P3, Medium)
diff --git a/planning/tasks/6fq9zy15j5pz-task-new-json-expose-the-minted-id-as-its-own-field.md b/planning/tasks/6fq9zy15j5pz-task-new-json-expose-the-minted-id-as-its-own-field.md
new file mode 100644
index 0000000..d601844
--- /dev/null
+++ b/planning/tasks/6fq9zy15j5pz-task-new-json-expose-the-minted-id-as-its-own-field.md
@@ -0,0 +1,31 @@
+---
+schema: 1
+id: 6fq9zy15j5pz
+status: ready-to-start
+epic: 20-cli-ux-and-ergonomics
+description: task new --json created.id is the slug; the minted id is only inside created.path.
+effort: Unknown
+tier: 3
+priority: low
+autonomy_level: 3
+tags: [json, id]
+created: "2026-07-18"
+---
+# task new --json: expose the minted id as its own field
+
+## Objective
+
+`task new --json` returns `created.id` = the **slug**; the real minted id is
+only embedded in `created.path` (`tasks/<minted-id>-<slug>.md`). Scripting
+`created.id` expecting the id yields the slug. Confirmed:
+`{"created":{"id":"zzz-probe","path":"tasks/6fq9xy66fehv-zzz-probe.md"}}`.
+
+## Acceptance criteria
+
+- [ ] The minted id is a distinct top-level field on the created object
+- [ ] `id` vs `slug` are unambiguous in the envelope (and consistent across new for task/epic/audit)
+
+## Notes
+
+- Loci: the create envelope in internal/wire (created DTO) + the CLI `new` handlers.
+- Source: https://github.com/andy-esch/taskflow/issues/105 (P4, Low)
diff --git a/planning/tasks/6fq9zy172apg-hard-delete-for-a-mistaken-create-task-rm.md b/planning/tasks/6fq9zy172apg-hard-delete-for-a-mistaken-create-task-rm.md
new file mode 100644
index 0000000..8210adb
--- /dev/null
+++ b/planning/tasks/6fq9zy172apg-hard-delete-for-a-mistaken-create-task-rm.md
@@ -0,0 +1,30 @@
+---
+schema: 1
+id: 6fq9zy172apg
+status: ready-to-start
+epic: 20-cli-ux-and-ergonomics
+description: No task rm/audit rm; undoing a mistaken create needs deprecate (tombstone) or a raw rm (bypasses the CLI).
+effort: Unknown
+tier: 3
+priority: low
+autonomy_level: 3
+tags: [lifecycle]
+created: "2026-07-18"
+---
+# Hard-delete for a mistaken create (task rm / audit rm)
+
+## Objective
+
+A duplicate or mistaken `task new` can only be undone with `deprecate` (leaves a
+tombstone) or by `rm`-ing the file (bypasses the CLI, against the "drive the
+lifecycle via tskflwctl" principle). There is no hard-delete verb. Confirmed: no
+`task rm`/`audit rm` today.
+
+## Acceptance criteria
+
+- [ ] A `task rm` (and `audit rm`) removes a never-committed mistake through the CLI
+- [ ] Guard-railed so it's clearly for mistakes, not a routine lifecycle move (vs deprecate)
+
+## Notes
+
+- Source: https://github.com/andy-esch/taskflow/issues/105 (P5, Low)