# Task AcceptanceCriteria Implementation Plan

> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.

**Goal:** Add `AcceptanceCriteria []string` to `task.Task`, mirroring `story.Story`'s existing pattern exactly, and expose it as a `spawn_subtask` parameter so a decomposing builder can state what each subtask must satisfy. This is piece 1 of `docs/superpowers/specs/2026-07-09-recursive-arbitrated-review-design.md`'s implementation order — additive, no behavior change to anything existing.

**Architecture:** `task.Task` gains one new field, stored as `acceptance_criteria_json` (same column-naming/JSON-marshal convention as `depends_on_json`/`tags_json`). Touching this column means touching every one of the 5 places `internal/storage/db.go` repeats the tasks table's `SELECT` column list verbatim — this plan extracts a `taskSelectColumns()` helper (mirroring the already-established `storySelectColumns()` in `internal/storage/story.go`) at the same time, so a future column addition only requires one edit, not five kept in sync by hand.

**Tech Stack:** Go, SQLite (additive `ALTER TABLE`, no data migration needed — new column defaults to `'[]'`).

## Global Constraints

- `AcceptanceCriteria` is **only settable at subtask creation** via `spawn_subtask` — it is intentionally not added to `storage.TaskUpdate`/`UpdateTask` (which already excludes several other `Task` fields, e.g. `ElaborationInput`, `ParentTaskID` — it is a curated subset for the "edit and reset to PENDING" REST use case, not an exhaustive field list).
- Do not touch anything about arbitrated review's trigger, `maybeUnblockParent`, or the story-level fix loop's `RootTaskID` mechanism — those are separate, later pieces of the same design spec's implementation order.

---

## Task 1: Storage layer

**Files:**
- Modify: `internal/task/task.go` (`Task` struct)
- Modify: `internal/storage/db.go` (schema, migration, new `taskSelectColumns()` helper, `CreateTask`, `GetTask`, `ListTasks`, `ListSubtasks`, `ListDependents`, `ResetTaskForRetry`, `scanTask`)
- Test: `internal/storage/db_test.go` (append)

**Interfaces:**
- Produces: `task.Task.AcceptanceCriteria []string`, round-tripped through `storage.DB.CreateTask`/`GetTask` (and every other read path). Task 2 consumes this field directly when wiring `spawn_subtask`.

- [ ] **Step 1: Add the field to `task.Task`**

In `internal/task/task.go`, find the `Task` struct's `DependsOn` field:

```go
	DependsOn   []string     `yaml:"depends_on" json:"depends_on"`
```

Add directly after it:

```go
	DependsOn   []string     `yaml:"depends_on" json:"depends_on"`
	// AcceptanceCriteria, when non-empty, states what this task's work must
	// satisfy — set by a decomposing parent via spawn_subtask's
	// acceptance_criteria parameter (mirrors story.Story.AcceptanceCriteria
	// exactly). Unused by execution itself; a later phase's arbitrated-review
	// generalization reads it when evaluating this task's work, falling back
	// to the enclosing story's own AcceptanceCriteria when empty.
	AcceptanceCriteria []string `yaml:"acceptance_criteria" json:"acceptance_criteria"`
```

- [ ] **Step 2: Write the failing storage test**

Append to `internal/storage/db_test.go`, directly after `TestCreateTask_Project_RoundTrip`:

```go
func TestCreateTask_AcceptanceCriteria_RoundTrip(t *testing.T) {
	db := testDB(t)
	now := time.Now().UTC().Truncate(time.Second)

	tk := &task.Task{
		ID:                 "ac-1",
		Name:               "Task With Criteria",
		Agent:              task.AgentConfig{Type: "claude", Instructions: "do it"},
		Priority:           task.PriorityNormal,
		Tags:               []string{},
		DependsOn:          []string{},
		AcceptanceCriteria: []string{"tests pass", "no new lint warnings"},
		Retry:              task.RetryConfig{MaxAttempts: 1, Backoff: "linear"},
		State:              task.StatePending,
		CreatedAt:          now,
		UpdatedAt:          now,
	}
	if err := db.CreateTask(tk); err != nil {
		t.Fatalf("creating task: %v", err)
	}

	got, err := db.GetTask("ac-1")
	if err != nil {
		t.Fatalf("getting task: %v", err)
	}
	if len(got.AcceptanceCriteria) != 2 || got.AcceptanceCriteria[0] != "tests pass" || got.AcceptanceCriteria[1] != "no new lint warnings" {
		t.Errorf("AcceptanceCriteria: want [tests pass, no new lint warnings], got %+v", got.AcceptanceCriteria)
	}
}

// TestCreateTask_AcceptanceCriteria_DefaultsEmpty proves a task created
// without AcceptanceCriteria round-trips as an empty (not nil) slice, the
// same "never null" convention story.Story.AcceptanceCriteria already uses.
func TestCreateTask_AcceptanceCriteria_DefaultsEmpty(t *testing.T) {
	db := testDB(t)
	now := time.Now().UTC().Truncate(time.Second)

	tk := &task.Task{
		ID:        "ac-2",
		Name:      "Task Without Criteria",
		Agent:     task.AgentConfig{Type: "claude", Instructions: "do it"},
		Priority:  task.PriorityNormal,
		Tags:      []string{},
		DependsOn: []string{},
		Retry:     task.RetryConfig{MaxAttempts: 1, Backoff: "linear"},
		State:     task.StatePending,
		CreatedAt: now,
		UpdatedAt: now,
	}
	if err := db.CreateTask(tk); err != nil {
		t.Fatalf("creating task: %v", err)
	}

	got, err := db.GetTask("ac-2")
	if err != nil {
		t.Fatalf("getting task: %v", err)
	}
	if got.AcceptanceCriteria == nil {
		t.Error("AcceptanceCriteria: want empty slice, got nil")
	}
	if len(got.AcceptanceCriteria) != 0 {
		t.Errorf("AcceptanceCriteria: want empty, got %+v", got.AcceptanceCriteria)
	}
}
```

- [ ] **Step 3: Run tests to verify they fail**

Run: `go test ./internal/storage/ -run TestCreateTask_AcceptanceCriteria -v`
Expected: compile failure — `task.Task` has no `AcceptanceCriteria` field yet (Step 1 alone doesn't make this pass; the storage layer doesn't read/write it yet). Paste the actual output.

- [ ] **Step 4: Add the column to the schema and migrations**

In `internal/storage/db.go`, find the `CREATE TABLE IF NOT EXISTS tasks` block:

```go
	CREATE TABLE IF NOT EXISTS tasks (
		id TEXT PRIMARY KEY,
		name TEXT NOT NULL,
		description TEXT,
		config_json TEXT NOT NULL,
		priority TEXT NOT NULL DEFAULT 'normal',
		timeout_ns INTEGER NOT NULL DEFAULT 0,
		retry_json TEXT NOT NULL DEFAULT '{}',
		tags_json TEXT NOT NULL DEFAULT '[]',
		depends_on_json TEXT NOT NULL DEFAULT '[]',
		parent_task_id TEXT,
		state TEXT NOT NULL DEFAULT 'PENDING',
		created_at DATETIME NOT NULL,
		updated_at DATETIME NOT NULL
	);
```

Replace it with:

```go
	CREATE TABLE IF NOT EXISTS tasks (
		id TEXT PRIMARY KEY,
		name TEXT NOT NULL,
		description TEXT,
		config_json TEXT NOT NULL,
		priority TEXT NOT NULL DEFAULT 'normal',
		timeout_ns INTEGER NOT NULL DEFAULT 0,
		retry_json TEXT NOT NULL DEFAULT '{}',
		tags_json TEXT NOT NULL DEFAULT '[]',
		depends_on_json TEXT NOT NULL DEFAULT '[]',
		acceptance_criteria_json TEXT NOT NULL DEFAULT '[]',
		parent_task_id TEXT,
		state TEXT NOT NULL DEFAULT 'PENDING',
		created_at DATETIME NOT NULL,
		updated_at DATETIME NOT NULL
	);
```

Find the last entry in the `migrations` slice:

```go
		`ALTER TABLE tasks ADD COLUMN needs_review BOOLEAN NOT NULL DEFAULT 0`,
	}
```

Replace it with:

```go
		`ALTER TABLE tasks ADD COLUMN needs_review BOOLEAN NOT NULL DEFAULT 0`,
		`ALTER TABLE tasks ADD COLUMN acceptance_criteria_json TEXT NOT NULL DEFAULT '[]'`,
	}
```

- [ ] **Step 5: Add `taskSelectColumns()` and replace all 5 duplicated SELECT column lists**

In `internal/storage/db.go`, find `GetTask`:

```go
// GetTask retrieves a task by ID.
func (s *DB) GetTask(id string) (*task.Task, error) {
	row := s.db.QueryRow(`SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review FROM tasks WHERE id = ?`, id)
	return scanTask(row)
}
```

Replace it with (adding the new helper directly above `GetTask`):

```go
// taskSelectColumns is the shared SELECT column list every task-reading
// query in this file uses — extracted so a future column addition is one
// edit instead of five kept in sync by hand. Mirrors storySelectColumns in
// internal/storage/story.go.
func taskSelectColumns() string {
	return `SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, acceptance_criteria_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review`
}

// GetTask retrieves a task by ID.
func (s *DB) GetTask(id string) (*task.Task, error) {
	row := s.db.QueryRow(taskSelectColumns()+` FROM tasks WHERE id = ?`, id)
	return scanTask(row)
}
```

In the same file, find `ListTasks`'s query assignment:

```go
	query := `SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review FROM tasks WHERE 1=1`
```

Replace it with:

```go
	query := taskSelectColumns() + ` FROM tasks WHERE 1=1`
```

Find `ListSubtasks`:

```go
// ListSubtasks returns all tasks whose parent_task_id matches the given ID.
func (s *DB) ListSubtasks(parentID string) ([]*task.Task, error) {
	rows, err := s.db.Query(`SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review FROM tasks WHERE parent_task_id = ? ORDER BY created_at ASC`, parentID)
```

Replace the query line with:

```go
	rows, err := s.db.Query(taskSelectColumns()+` FROM tasks WHERE parent_task_id = ? ORDER BY created_at ASC`, parentID)
```

Find `ListDependents`:

```go
	rows, err := s.db.Query(`SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review FROM tasks`)
```

Replace it with:

```go
	rows, err := s.db.Query(taskSelectColumns() + ` FROM tasks`)
```

Find `ResetTaskForRetry`:

```go
	t, err := scanTask(tx.QueryRow(`SELECT id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at, rejection_comment, question_json, summary, interactions_json, needs_review FROM tasks WHERE id = ?`, id))
```

Replace it with:

```go
	t, err := scanTask(tx.QueryRow(taskSelectColumns()+` FROM tasks WHERE id = ?`, id))
```

- [ ] **Step 6: Update `CreateTask` to write the new column**

In the same file, find `CreateTask`:

```go
func (s *DB) CreateTask(t *task.Task) error {
	configJSON, err := json.Marshal(t.Agent)
	if err != nil {
		return fmt.Errorf("marshaling config: %w", err)
	}
	retryJSON, err := json.Marshal(t.Retry)
	if err != nil {
		return fmt.Errorf("marshaling retry: %w", err)
	}
	tagsJSON, err := json.Marshal(t.Tags)
	if err != nil {
		return fmt.Errorf("marshaling tags: %w", err)
	}
	depsJSON, err := json.Marshal(t.DependsOn)
	if err != nil {
		return fmt.Errorf("marshaling depends_on: %w", err)
	}

	tx, err := s.db.Begin()
	if err != nil {
		return err
	}
	defer tx.Rollback() //nolint:errcheck

	if _, err = tx.Exec(`
		INSERT INTO tasks (id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, parent_task_id, state, created_at, updated_at)
		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
		t.ID, t.Name, t.Description, t.ElaborationInput, t.Project, t.RepositoryURL, string(configJSON), string(t.Priority),
		t.Timeout.Duration.Nanoseconds(), string(retryJSON), string(tagsJSON), string(depsJSON),
		t.ParentTaskID, string(t.State), t.CreatedAt.UTC(), t.UpdatedAt.UTC(),
	); err != nil {
		return err
	}
```

Replace it with:

```go
func (s *DB) CreateTask(t *task.Task) error {
	configJSON, err := json.Marshal(t.Agent)
	if err != nil {
		return fmt.Errorf("marshaling config: %w", err)
	}
	retryJSON, err := json.Marshal(t.Retry)
	if err != nil {
		return fmt.Errorf("marshaling retry: %w", err)
	}
	tagsJSON, err := json.Marshal(t.Tags)
	if err != nil {
		return fmt.Errorf("marshaling tags: %w", err)
	}
	depsJSON, err := json.Marshal(t.DependsOn)
	if err != nil {
		return fmt.Errorf("marshaling depends_on: %w", err)
	}
	acceptanceCriteria := t.AcceptanceCriteria
	if acceptanceCriteria == nil {
		acceptanceCriteria = []string{}
	}
	acceptanceCriteriaJSON, err := json.Marshal(acceptanceCriteria)
	if err != nil {
		return fmt.Errorf("marshaling acceptance_criteria: %w", err)
	}

	tx, err := s.db.Begin()
	if err != nil {
		return err
	}
	defer tx.Rollback() //nolint:errcheck

	if _, err = tx.Exec(`
		INSERT INTO tasks (id, name, description, elaboration_input, project, repository_url, config_json, priority, timeout_ns, retry_json, tags_json, depends_on_json, acceptance_criteria_json, parent_task_id, state, created_at, updated_at)
		VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
		t.ID, t.Name, t.Description, t.ElaborationInput, t.Project, t.RepositoryURL, string(configJSON), string(t.Priority),
		t.Timeout.Duration.Nanoseconds(), string(retryJSON), string(tagsJSON), string(depsJSON), string(acceptanceCriteriaJSON),
		t.ParentTaskID, string(t.State), t.CreatedAt.UTC(), t.UpdatedAt.UTC(),
	); err != nil {
		return err
	}
```

- [ ] **Step 7: Update `scanTask` to read the new column**

In the same file, find `scanTask`:

```go
func scanTask(row scanner) (*task.Task, error) {
	var (
		t                  task.Task
		configJSON         string
		retryJSON          string
		tagsJSON           string
		depsJSON           string
		state              string
		priority           string
		timeoutNS          int64
		parentTaskID       sql.NullString
		elaborationInput   sql.NullString
		project            sql.NullString
		repositoryURL      sql.NullString
		rejectionComment   sql.NullString
		questionJSON       sql.NullString
		summary            sql.NullString
		interactionsJSON   sql.NullString
	)
	err := row.Scan(
		&t.ID, &t.Name, &t.Description, &elaborationInput, &project, &repositoryURL,
		&configJSON, &priority, &timeoutNS, &retryJSON, &tagsJSON, &depsJSON,
		&parentTaskID, &state, &t.CreatedAt, &t.UpdatedAt,
		&rejectionComment, &questionJSON, &summary, &interactionsJSON, &t.NeedsReview,
	)
	t.ParentTaskID = parentTaskID.String
	t.ElaborationInput = elaborationInput.String
	t.Project = project.String
	t.RepositoryURL = repositoryURL.String
	t.RejectionComment = rejectionComment.String
	t.QuestionJSON = questionJSON.String
	t.Summary = summary.String
	if err != nil {
		return nil, err
	}
	t.State = task.State(state)
	t.Priority = task.Priority(priority)
	t.Timeout.Duration = time.Duration(timeoutNS)
	if err := json.Unmarshal([]byte(configJSON), &t.Agent); err != nil {
		return nil, fmt.Errorf("unmarshaling agent config: %w", err)
	}
	if err := json.Unmarshal([]byte(retryJSON), &t.Retry); err != nil {
		return nil, fmt.Errorf("unmarshaling retry: %w", err)
	}
	if err := json.Unmarshal([]byte(tagsJSON), &t.Tags); err != nil {
		return nil, fmt.Errorf("unmarshaling tags: %w", err)
	}
	if err := json.Unmarshal([]byte(depsJSON), &t.DependsOn); err != nil {
		return nil, fmt.Errorf("unmarshaling depends_on: %w", err)
	}
	raw := interactionsJSON.String
	if raw == "" {
		raw = "[]"
	}
	if err := json.Unmarshal([]byte(raw), &t.Interactions); err != nil {
		return nil, fmt.Errorf("unmarshaling interactions: %w", err)
	}
	return &t, nil
}
```

Replace it with:

```go
func scanTask(row scanner) (*task.Task, error) {
	var (
		t                      task.Task
		configJSON             string
		retryJSON              string
		tagsJSON               string
		depsJSON               string
		acceptanceCriteriaJSON string
		state                  string
		priority               string
		timeoutNS              int64
		parentTaskID           sql.NullString
		elaborationInput       sql.NullString
		project                sql.NullString
		repositoryURL          sql.NullString
		rejectionComment       sql.NullString
		questionJSON           sql.NullString
		summary                sql.NullString
		interactionsJSON       sql.NullString
	)
	err := row.Scan(
		&t.ID, &t.Name, &t.Description, &elaborationInput, &project, &repositoryURL,
		&configJSON, &priority, &timeoutNS, &retryJSON, &tagsJSON, &depsJSON, &acceptanceCriteriaJSON,
		&parentTaskID, &state, &t.CreatedAt, &t.UpdatedAt,
		&rejectionComment, &questionJSON, &summary, &interactionsJSON, &t.NeedsReview,
	)
	t.ParentTaskID = parentTaskID.String
	t.ElaborationInput = elaborationInput.String
	t.Project = project.String
	t.RepositoryURL = repositoryURL.String
	t.RejectionComment = rejectionComment.String
	t.QuestionJSON = questionJSON.String
	t.Summary = summary.String
	if err != nil {
		return nil, err
	}
	t.State = task.State(state)
	t.Priority = task.Priority(priority)
	t.Timeout.Duration = time.Duration(timeoutNS)
	if err := json.Unmarshal([]byte(configJSON), &t.Agent); err != nil {
		return nil, fmt.Errorf("unmarshaling agent config: %w", err)
	}
	if err := json.Unmarshal([]byte(retryJSON), &t.Retry); err != nil {
		return nil, fmt.Errorf("unmarshaling retry: %w", err)
	}
	if err := json.Unmarshal([]byte(tagsJSON), &t.Tags); err != nil {
		return nil, fmt.Errorf("unmarshaling tags: %w", err)
	}
	if err := json.Unmarshal([]byte(depsJSON), &t.DependsOn); err != nil {
		return nil, fmt.Errorf("unmarshaling depends_on: %w", err)
	}
	acRaw := acceptanceCriteriaJSON
	if acRaw == "" {
		acRaw = "[]"
	}
	if err := json.Unmarshal([]byte(acRaw), &t.AcceptanceCriteria); err != nil {
		return nil, fmt.Errorf("unmarshaling acceptance_criteria: %w", err)
	}
	raw := interactionsJSON.String
	if raw == "" {
		raw = "[]"
	}
	if err := json.Unmarshal([]byte(raw), &t.Interactions); err != nil {
		return nil, fmt.Errorf("unmarshaling interactions: %w", err)
	}
	return &t, nil
}
```

- [ ] **Step 8: Run tests to verify they pass**

Run: `go test ./internal/storage/ -run TestCreateTask -v`
Expected: PASS, all `TestCreateTask_*` tests, including the two new ones and every pre-existing one (`TestCreateTask_AndGetTask`, `TestCreateTask_Project_RoundTrip`, `TestCreateTask_Subtask_EmitsSubtaskSpawnedOnParent`, `TestCreateTask_TopLevel_EmitsNoEvent`) — this proves `taskSelectColumns()`'s column order matches `scanTask`'s scan order exactly, and that no existing behavior broke. Paste the actual output.

- [ ] **Step 9: Run the full package suite, then the full repo suite**

Run: `go test ./internal/storage/...`
Expected: PASS for everything — `ListTasks`/`ListSubtasks`/`ListDependents`/`ResetTaskForRetry` all have their own existing tests elsewhere in this package; if `taskSelectColumns()`'s column list or `scanTask`'s scan order is wrong, one of them will fail with a scan-type-mismatch error, not silently return wrong data.

Then run: `go test ./...` (the entire repo). This must also pass — paste the actual output. If `internal/api` flakes under the race detector on a one-off "sql: database is closed" teardown error unrelated to this task's actual changes, rerun once before treating it as real.

- [ ] **Step 10: Commit and push to `main` directly, matching this repo's existing workflow**

```bash
git add internal/task/task.go internal/storage/db.go internal/storage/db_test.go
git commit -m "feat(task,storage): add Task.AcceptanceCriteria, mirroring story.Story's pattern"
```

## Mandatory verification disclosure

When you call report_summary, paste the actual terminal output of every test command above — literal pass/fail counts, not a claim of success, including the full-repo `go test ./...` run. If anything here is ambiguous or conflicts with what you find in the actual repo, call ask_user and describe the specific conflict — don't guess or silently decide.

---

## Task 2: Expose `acceptance_criteria` on `spawn_subtask`

**Files:**
- Modify: `internal/agentchannel/agentchannel.go` (`SubtaskSpec`)
- Modify: `internal/executor/channel.go` (`storeChannel.SpawnSubtask`)
- Modify: `internal/agentloop/tools.go` (schema + case handler)
- Modify: `internal/executor/agentmcp.go` (input struct + MCP tool)
- Test: `internal/executor/channel_test.go` (append)

**Interfaces:**
- Consumes: `task.Task.AcceptanceCriteria` (Task 1).
- Produces: nothing new for later work — this closes the `spawn_subtask` half of this plan's scope. The arbitrated-review generalization that actually *reads* a task's `AcceptanceCriteria` when evaluating it is a later, separate piece of the design spec's implementation order.

- [ ] **Step 1: Add `AcceptanceCriteria` to `SubtaskSpec`**

In `internal/agentchannel/agentchannel.go`, find the end of `SubtaskSpec`:

```go
	// DependsOn, when non-empty, names sibling task IDs (returned by a prior
	// SpawnSubtask call in the same decomposition) this new subtask must wait
	// for — set directly on the created child's task.DependsOn, so the
	// existing dispatch-gating and cascade-fail-on-dependency-failure
	// machinery (already built for top-level tasks) applies unchanged. Empty
	// (the default) preserves today's behavior: every spawned subtask is
	// structurally independent/parallel.
	DependsOn []string
}
```

Replace it with:

```go
	// DependsOn, when non-empty, names sibling task IDs (returned by a prior
	// SpawnSubtask call in the same decomposition) this new subtask must wait
	// for — set directly on the created child's task.DependsOn, so the
	// existing dispatch-gating and cascade-fail-on-dependency-failure
	// machinery (already built for top-level tasks) applies unchanged. Empty
	// (the default) preserves today's behavior: every spawned subtask is
	// structurally independent/parallel.
	DependsOn []string
	// AcceptanceCriteria, when non-empty, states what this subtask's work
	// must satisfy — set directly on the created child's
	// task.Task.AcceptanceCriteria. Empty (the default) leaves the child with
	// no criteria of its own; a later phase's arbitrated-review
	// generalization falls back to the enclosing story's AcceptanceCriteria
	// in that case.
	AcceptanceCriteria []string
}
```

- [ ] **Step 2: Wire it through `storeChannel.SpawnSubtask`**

In `internal/executor/channel.go`, find:

```go
	child := &task.Task{
		ID:           uuid.NewString(),
		Name:         spec.Name,
		ParentTaskID: c.taskID,
		Agent:        agent,
		Priority:     task.PriorityNormal,
		DependsOn:    spec.DependsOn,
		State:        task.StatePending,
		CreatedAt:    now,
		UpdatedAt:    now,
	}
```

Replace it with:

```go
	child := &task.Task{
		ID:                 uuid.NewString(),
		Name:                spec.Name,
		ParentTaskID:        c.taskID,
		Agent:               agent,
		Priority:            task.PriorityNormal,
		DependsOn:           spec.DependsOn,
		AcceptanceCriteria:  spec.AcceptanceCriteria,
		State:               task.StatePending,
		CreatedAt:           now,
		UpdatedAt:           now,
	}
```

- [ ] **Step 3: Write the failing tests**

Append to `internal/executor/channel_test.go`, directly after `TestStoreChannel_SpawnSubtask_DependsOnDefaultsEmpty`:

```go
// TestStoreChannel_SpawnSubtask_SetsAcceptanceCriteria proves a caller can
// state what a spawned subtask's work must satisfy.
func TestStoreChannel_SpawnSubtask_SetsAcceptanceCriteria(t *testing.T) {
	store := &fakeChannelStore{}
	ch := newStoreChannel(store, "parent-task-1")

	id, err := ch.SpawnSubtask(context.Background(), agentchannel.SubtaskSpec{
		Name:               "step 1",
		Instructions:       "do the thing",
		AcceptanceCriteria: []string{"tests pass", "no new lint warnings"},
	})
	if err != nil {
		t.Fatalf("spawn subtask: %v", err)
	}

	var got *task.Task
	for _, ct := range store.createdTasks {
		if ct.ID == id {
			got = ct
		}
	}
	if got == nil {
		t.Fatal("subtask not found in store.createdTasks")
	}
	if len(got.AcceptanceCriteria) != 2 || got.AcceptanceCriteria[0] != "tests pass" || got.AcceptanceCriteria[1] != "no new lint warnings" {
		t.Errorf("AcceptanceCriteria: want [tests pass, no new lint warnings], got %+v", got.AcceptanceCriteria)
	}
}

// TestStoreChannel_SpawnSubtask_AcceptanceCriteriaDefaultsEmpty proves the
// backward-compatible default: a caller that never sets AcceptanceCriteria
// (every pre-existing caller) gets a child task with none, exactly as before
// this change.
func TestStoreChannel_SpawnSubtask_AcceptanceCriteriaDefaultsEmpty(t *testing.T) {
	store := &fakeChannelStore{}
	ch := newStoreChannel(store, "parent-task-1")

	id, err := ch.SpawnSubtask(context.Background(), agentchannel.SubtaskSpec{
		Name:         "solo step",
		Instructions: "do the thing",
	})
	if err != nil {
		t.Fatalf("spawn subtask: %v", err)
	}

	var got *task.Task
	for _, ct := range store.createdTasks {
		if ct.ID == id {
			got = ct
		}
	}
	if got == nil {
		t.Fatal("subtask not found in store.createdTasks")
	}
	if len(got.AcceptanceCriteria) != 0 {
		t.Errorf("expected empty AcceptanceCriteria by default, got %v", got.AcceptanceCriteria)
	}
}
```

- [ ] **Step 4: Run tests to verify they fail, then implement, then verify they pass**

Run: `go test ./internal/executor/ -run TestStoreChannel_SpawnSubtask_.*AcceptanceCriteria -v`
Expected: FAIL (compile error, `SubtaskSpec` has no `AcceptanceCriteria` field) — this is the point where you actually make Steps 1-2's edits above if you haven't yet, then re-run and expect PASS. Paste both the failing and passing output.

- [ ] **Step 5: Add the tool definition and case handler to the native tool-use loop**

In `internal/agentloop/tools.go`, find the `spawn_subtask` tool definition:

```go
		{
			Name:        "spawn_subtask",
			Description: "Create a child task to be executed separately. Use this to break large work into focused pieces.",
			ParametersJSONSchema: map[string]any{
				"type": "object",
				"properties": map[string]any{
					"name":           strProp("short descriptive name for the subtask"),
					"instructions":   strProp("complete instructions for the subtask agent"),
					"model":          strProp("optional model override"),
					"max_budget_usd": map[string]any{"type": "number", "description": "optional budget cap in USD"},
					"role":           strProp("optional role name to dispatch the subtask through instead of a fixed model (e.g. an evaluator role); when set, model is ignored and the role's escalation ladder picks the provider/model"),
					"depends_on":     map[string]any{"type": "array", "items": map[string]any{"type": "string"}, "description": "optional list of sibling subtask IDs (returned by prior spawn_subtask calls in this same decomposition) this subtask must wait for before it can run"},
				},
				"required": []string{"name", "instructions"},
			},
		},
```

Replace it with:

```go
		{
			Name:        "spawn_subtask",
			Description: "Create a child task to be executed separately. Use this to break large work into focused pieces.",
			ParametersJSONSchema: map[string]any{
				"type": "object",
				"properties": map[string]any{
					"name":                strProp("short descriptive name for the subtask"),
					"instructions":        strProp("complete instructions for the subtask agent"),
					"model":               strProp("optional model override"),
					"max_budget_usd":      map[string]any{"type": "number", "description": "optional budget cap in USD"},
					"role":                strProp("optional role name to dispatch the subtask through instead of a fixed model (e.g. an evaluator role); when set, model is ignored and the role's escalation ladder picks the provider/model"),
					"depends_on":          map[string]any{"type": "array", "items": map[string]any{"type": "string"}, "description": "optional list of sibling subtask IDs (returned by prior spawn_subtask calls in this same decomposition) this subtask must wait for before it can run"},
					"acceptance_criteria": map[string]any{"type": "array", "items": map[string]any{"type": "string"}, "description": "optional list of concrete criteria this subtask's work must satisfy"},
				},
				"required": []string{"name", "instructions"},
			},
		},
```

In the same file, find the `spawn_subtask` case handler:

```go
	case "spawn_subtask":
		var a struct {
			Name         string   `json:"name"`
			Instructions string   `json:"instructions"`
			Model        string   `json:"model"`
			MaxBudgetUSD float64  `json:"max_budget_usd"`
			Role         string   `json:"role"`
			DependsOn    []string `json:"depends_on"`
		}
		_ = json.Unmarshal([]byte(argsJSON), &a)
		id, ssErr := l.Channel.SpawnSubtask(ctx, agentchannel.SubtaskSpec{
			Name:         a.Name,
			Instructions: a.Instructions,
			Model:        a.Model,
			MaxBudgetUSD: a.MaxBudgetUSD,
			Role:         a.Role,
			DependsOn:    a.DependsOn,
		})
```

Replace it with:

```go
	case "spawn_subtask":
		var a struct {
			Name               string   `json:"name"`
			Instructions       string   `json:"instructions"`
			Model              string   `json:"model"`
			MaxBudgetUSD       float64  `json:"max_budget_usd"`
			Role               string   `json:"role"`
			DependsOn          []string `json:"depends_on"`
			AcceptanceCriteria []string `json:"acceptance_criteria"`
		}
		_ = json.Unmarshal([]byte(argsJSON), &a)
		id, ssErr := l.Channel.SpawnSubtask(ctx, agentchannel.SubtaskSpec{
			Name:               a.Name,
			Instructions:       a.Instructions,
			Model:               a.Model,
			MaxBudgetUSD:        a.MaxBudgetUSD,
			Role:                a.Role,
			DependsOn:           a.DependsOn,
			AcceptanceCriteria:  a.AcceptanceCriteria,
		})
```

- [ ] **Step 6: Add the MCP input field and wire the registration**

In `internal/executor/agentmcp.go`, find `spawnSubtaskInput`:

```go
type spawnSubtaskInput struct {
	Name         string   `json:"name" jsonschema:"short descriptive name for the subtask"`
	Instructions string   `json:"instructions" jsonschema:"complete instructions for the subtask agent"`
	Model        string   `json:"model,omitempty" jsonschema:"optional model override, e.g. sonnet or opus"`
	MaxBudgetUSD float64  `json:"max_budget_usd,omitempty" jsonschema:"optional budget cap in USD"`
	Role         string   `json:"role,omitempty" jsonschema:"optional role name to dispatch the subtask through instead of a fixed model (e.g. an evaluator role); when set, model is ignored and the role's escalation ladder picks the provider/model"`
	DependsOn    []string `json:"depends_on,omitempty" jsonschema:"optional list of sibling subtask IDs (returned by prior spawn_subtask calls in this same decomposition) this subtask must wait for before it can run"`
}
```

Replace it with:

```go
type spawnSubtaskInput struct {
	Name               string   `json:"name" jsonschema:"short descriptive name for the subtask"`
	Instructions       string   `json:"instructions" jsonschema:"complete instructions for the subtask agent"`
	Model              string   `json:"model,omitempty" jsonschema:"optional model override, e.g. sonnet or opus"`
	MaxBudgetUSD       float64  `json:"max_budget_usd,omitempty" jsonschema:"optional budget cap in USD"`
	Role               string   `json:"role,omitempty" jsonschema:"optional role name to dispatch the subtask through instead of a fixed model (e.g. an evaluator role); when set, model is ignored and the role's escalation ladder picks the provider/model"`
	DependsOn          []string `json:"depends_on,omitempty" jsonschema:"optional list of sibling subtask IDs (returned by prior spawn_subtask calls in this same decomposition) this subtask must wait for before it can run"`
	AcceptanceCriteria []string `json:"acceptance_criteria,omitempty" jsonschema:"optional list of concrete criteria this subtask's work must satisfy"`
}
```

In the same file, find the `spawn_subtask` `mcp.AddTool` call:

```go
	mcp.AddTool(s, &mcp.Tool{
		Name:        "spawn_subtask",
		Description: "Create a child task to be executed separately. Use this to break large work into focused pieces, then finish your turn.",
	}, func(ctx context.Context, _ *mcp.CallToolRequest, in spawnSubtaskInput) (*mcp.CallToolResult, any, error) {
		id, err := ch.SpawnSubtask(ctx, SubtaskSpec{
			Name:         in.Name,
			Instructions: in.Instructions,
			Model:        in.Model,
			MaxBudgetUSD: in.MaxBudgetUSD,
			Role:         in.Role,
			DependsOn:    in.DependsOn,
		})
		if err != nil {
			return nil, nil, err
		}
		return textResult("Created subtask " + id), nil, nil
	})
```

Replace it with:

```go
	mcp.AddTool(s, &mcp.Tool{
		Name:        "spawn_subtask",
		Description: "Create a child task to be executed separately. Use this to break large work into focused pieces, then finish your turn.",
	}, func(ctx context.Context, _ *mcp.CallToolRequest, in spawnSubtaskInput) (*mcp.CallToolResult, any, error) {
		id, err := ch.SpawnSubtask(ctx, SubtaskSpec{
			Name:                in.Name,
			Instructions:        in.Instructions,
			Model:               in.Model,
			MaxBudgetUSD:        in.MaxBudgetUSD,
			Role:                in.Role,
			DependsOn:           in.DependsOn,
			AcceptanceCriteria:  in.AcceptanceCriteria,
		})
		if err != nil {
			return nil, nil, err
		}
		return textResult("Created subtask " + id), nil, nil
	})
```

- [ ] **Step 7: Run tests to verify they pass**

Run: `go test ./internal/agentloop/... ./internal/executor/...`
Expected: PASS for everything, no new tests needed for `tools.go`/`agentmcp.go` themselves (the underlying `SpawnSubtask` behavior is already covered by Task 2's own `channel_test.go` tests — this mirrors how `depends_on`'s exposure on both transports needed no new tests of its own in the earlier plan that added it).

- [ ] **Step 8: Run the full package suite, then the full repo suite**

Run: `go test ./internal/agentchannel/... ./internal/executor/... ./internal/agentloop/...`
Expected: PASS for everything.

Then run: `go test ./...` (the entire repo). This must also pass before you commit — paste the actual output. If `internal/api` flakes under the race detector on a one-off "sql: database is closed" teardown error unrelated to this task's actual changes, rerun once before treating it as real.

- [ ] **Step 9: Commit and push to `main` directly, matching this repo's existing workflow**

```bash
git add internal/agentchannel/agentchannel.go internal/executor/channel.go internal/executor/channel_test.go internal/agentloop/tools.go internal/executor/agentmcp.go
git commit -m "feat(agentchannel,executor,agentloop): expose acceptance_criteria on spawn_subtask"
```

## Mandatory verification disclosure

When you call report_summary, paste the actual terminal output of every test command above — literal pass/fail counts, not a claim of success, including the full-repo `go test ./...` run. If anything here is ambiguous or conflicts with what you find in the actual repo, call ask_user and describe the specific conflict — don't guess or silently decide.

---

## Final Verification

- [ ] Run `go build ./...` — passes.
- [ ] Run `go test ./...` — passes, full repo.
- [ ] Run `grep -rn "AcceptanceCriteria" internal/` to visually confirm every file in both tasks' File Maps was actually touched.