refactor(executor): update runners and tests for generic agents

4 files changed, 48 insertions, 37 deletions

diff --git a/internal/executor/claude.go b/internal/executor/claude.go
index b97f202..86a2ba5 100644
--- a/internal/executor/claude.go
+++ b/internal/executor/claude.go
@@ -56,9 +56,9 @@ func (r *ClaudeRunner) binaryPath() string {
 // It retries up to 3 times on rate-limit errors using exponential backoff.
 // If the agent writes a question file and exits, Run returns *BlockedError.
 func (r *ClaudeRunner) Run(ctx context.Context, t *task.Task, e *storage.Execution) error {
-	if t.Claude.WorkingDir != "" {
-		if _, err := os.Stat(t.Claude.WorkingDir); err != nil {
-			return fmt.Errorf("working_dir %q: %w", t.Claude.WorkingDir, err)
+	if t.Agent.WorkingDir != "" {
+		if _, err := os.Stat(t.Agent.WorkingDir); err != nil {
+			return fmt.Errorf("working_dir %q: %w", t.Agent.WorkingDir, err)
 		}
 	}
 
@@ -95,7 +95,7 @@ func (r *ClaudeRunner) Run(ctx context.Context, t *task.Task, e *storage.Executi
 			)
 		}
 		attempt++
-		return r.execOnce(ctx, args, t.Claude.WorkingDir, e)
+		return r.execOnce(ctx, args, t.Agent.WorkingDir, e)
 	})
 	if err != nil {
 		return err
@@ -208,21 +208,21 @@ func (r *ClaudeRunner) buildArgs(t *task.Task, e *storage.Execution, questionFil
 			"--output-format", "stream-json",
 			"--verbose",
 		}
-		permMode := t.Claude.PermissionMode
+		permMode := t.Agent.PermissionMode
 		if permMode == "" {
 			permMode = "bypassPermissions"
 		}
 		args = append(args, "--permission-mode", permMode)
-		if t.Claude.Model != "" {
-			args = append(args, "--model", t.Claude.Model)
+		if t.Agent.Model != "" {
+			args = append(args, "--model", t.Agent.Model)
 		}
 		return args
 	}
 
-	instructions := t.Claude.Instructions
-	allowedTools := t.Claude.AllowedTools
+	instructions := t.Agent.Instructions
+	allowedTools := t.Agent.AllowedTools
 
-	if !t.Claude.SkipPlanning {
+	if !t.Agent.SkipPlanning {
 		instructions = withPlanningPreamble(instructions)
 		// Ensure Bash is available so the agent can POST subtasks and ask questions.
 		hasBash := false
@@ -244,33 +244,33 @@ func (r *ClaudeRunner) buildArgs(t *task.Task, e *storage.Execution, questionFil
 		"--verbose",
 	}
 
-	if t.Claude.Model != "" {
-		args = append(args, "--model", t.Claude.Model)
+	if t.Agent.Model != "" {
+		args = append(args, "--model", t.Agent.Model)
 	}
-	if t.Claude.MaxBudgetUSD > 0 {
-		args = append(args, "--max-budget-usd", fmt.Sprintf("%.2f", t.Claude.MaxBudgetUSD))
+	if t.Agent.MaxBudgetUSD > 0 {
+		args = append(args, "--max-budget-usd", fmt.Sprintf("%.2f", t.Agent.MaxBudgetUSD))
 	}
 	// Default to bypassPermissions — claudomator runs tasks unattended, so
 	// prompting for write access would always stall execution. Tasks that need
 	// a more restrictive mode can set permission_mode explicitly.
-	permMode := t.Claude.PermissionMode
+	permMode := t.Agent.PermissionMode
 	if permMode == "" {
 		permMode = "bypassPermissions"
 	}
 	args = append(args, "--permission-mode", permMode)
-	if t.Claude.SystemPromptAppend != "" {
-		args = append(args, "--append-system-prompt", t.Claude.SystemPromptAppend)
+	if t.Agent.SystemPromptAppend != "" {
+		args = append(args, "--append-system-prompt", t.Agent.SystemPromptAppend)
 	}
 	for _, tool := range allowedTools {
 		args = append(args, "--allowedTools", tool)
 	}
-	for _, tool := range t.Claude.DisallowedTools {
+	for _, tool := range t.Agent.DisallowedTools {
 		args = append(args, "--disallowedTools", tool)
 	}
-	for _, f := range t.Claude.ContextFiles {
+	for _, f := range t.Agent.ContextFiles {
 		args = append(args, "--add-dir", f)
 	}
-	args = append(args, t.Claude.AdditionalArgs...)
+	args = append(args, t.Agent.AdditionalArgs...)
 
 	return args
 }
diff --git a/internal/executor/executor_test.go b/internal/executor/executor_test.go
index 6d13873..2f205f8 100644
--- a/internal/executor/executor_test.go
+++ b/internal/executor/executor_test.go
@@ -64,7 +64,7 @@ func makeTask(id string) *task.Task {
 	now := time.Now().UTC()
 	return &task.Task{
 		ID: id, Name: "Test " + id,
-		Claude:    task.ClaudeConfig{Instructions: "test"},
+		Agent:    task.AgentConfig{Type: "claude", Instructions: "test"},
 		Priority:  task.PriorityNormal,
 		Retry:     task.RetryConfig{MaxAttempts: 1, Backoff: "linear"},
 		Tags:      []string{},
@@ -77,8 +77,9 @@ func makeTask(id string) *task.Task {
 func TestPool_Submit_TopLevel_GoesToReady(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("ps-1") // no ParentTaskID → top-level
 	store.CreateTask(tk)
@@ -104,8 +105,9 @@ func TestPool_Submit_TopLevel_GoesToReady(t *testing.T) {
 func TestPool_Submit_Subtask_GoesToCompleted(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("sub-1")
 	tk.ParentTaskID = "parent-99" // subtask
@@ -132,8 +134,9 @@ func TestPool_Submit_Subtask_GoesToCompleted(t *testing.T) {
 func TestPool_Submit_Failure(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{err: fmt.Errorf("boom"), exitCode: 1}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("pf-1")
 	store.CreateTask(tk)
@@ -151,8 +154,9 @@ func TestPool_Submit_Failure(t *testing.T) {
 func TestPool_Submit_Timeout(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{delay: 5 * time.Second}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("pt-1")
 	tk.Timeout.Duration = 50 * time.Millisecond
@@ -168,8 +172,9 @@ func TestPool_Submit_Timeout(t *testing.T) {
 func TestPool_Submit_Cancellation(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{delay: 5 * time.Second}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	ctx, cancel := context.WithCancel(context.Background())
 	tk := makeTask("pc-1")
@@ -188,8 +193,9 @@ func TestPool_Submit_Cancellation(t *testing.T) {
 func TestPool_Cancel_StopsRunningTask(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{delay: 5 * time.Second}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("cancel-1")
 	store.CreateTask(tk)
@@ -209,8 +215,9 @@ func TestPool_Cancel_StopsRunningTask(t *testing.T) {
 func TestPool_Cancel_UnknownTask_ReturnsFalse(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	if ok := pool.Cancel("nonexistent"); ok {
 		t.Error("Cancel returned true for unknown task")
@@ -220,8 +227,9 @@ func TestPool_Cancel_UnknownTask_ReturnsFalse(t *testing.T) {
 func TestPool_AtCapacity(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{delay: time.Second}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(1, runner, store, logger)
+	pool := NewPool(1, runners, store, logger)
 
 	tk1 := makeTask("cap-1")
 	store.CreateTask(tk1)
@@ -265,8 +273,9 @@ func (m *logPatherMockRunner) Run(ctx context.Context, t *task.Task, e *storage.
 func TestPool_Execute_LogPathsPreSetBeforeRun(t *testing.T) {
 	store := testStore(t)
 	runner := &logPatherMockRunner{logDir: t.TempDir()}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("lp-1")
 	store.CreateTask(tk)
@@ -296,8 +305,9 @@ func TestPool_Execute_LogPathsPreSetBeforeRun(t *testing.T) {
 func TestPool_Execute_NoLogPather_PathsEmptyBeforeRun(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{} // does NOT implement LogPather
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(2, runner, store, logger)
+	pool := NewPool(2, runners, store, logger)
 
 	tk := makeTask("nolp-1")
 	store.CreateTask(tk)
@@ -313,8 +323,9 @@ func TestPool_Execute_NoLogPather_PathsEmptyBeforeRun(t *testing.T) {
 func TestPool_ConcurrentExecution(t *testing.T) {
 	store := testStore(t)
 	runner := &mockRunner{delay: 50 * time.Millisecond}
+	runners := map[string]Runner{"claude": runner}
 	logger := slog.New(slog.NewTextHandler(os.Stderr, &slog.HandlerOptions{Level: slog.LevelError}))
-	pool := NewPool(3, runner, store, logger)
+	pool := NewPool(3, runners, store, logger)
 
 	for i := 0; i < 3; i++ {
 		tk := makeTask(fmt.Sprintf("cc-%d", i))
diff --git a/internal/executor/gemini.go b/internal/executor/gemini.go
index e4bd50e..3cabed5 100644
--- a/internal/executor/gemini.go
+++ b/internal/executor/gemini.go
@@ -10,7 +10,6 @@ import (
 	"strings"
 	"sync"
 	"syscall"
-	"time"
 
 	"github.com/thepeterstone/claudomator/internal/storage"
 	"github.com/thepeterstone/claudomator/internal/task"
diff --git a/internal/executor/preamble.go b/internal/executor/preamble.go
index 71f8233..b20f7ea 100644
--- a/internal/executor/preamble.go
+++ b/internal/executor/preamble.go
@@ -23,11 +23,12 @@ Before doing any implementation work:
 
 1. Estimate: will this task take more than 3 minutes of implementation effort?
 
-2. If YES — break it down into subtasks:
-   - Create 3–7 discrete subtasks using the claudomator CLI, for example:
-       claudomator create "Subtask name" --instructions "..." --working-dir "/path" --parent-id "$CLAUDOMATOR_TASK_ID" --server "$CLAUDOMATOR_API_URL"
-   - Do NOT pass --start. Tasks will be queued and started in order by the operator.
+2. If YES — break it down:
+   - Create 3–7 discrete subtasks by POSTing to $CLAUDOMATOR_API_URL/api/tasks
+   - Each subtask POST body should be JSON with: name, agent.instructions, agent.working_dir (copy from current task), agent.model, agent.allowed_tools, and agent.skip_planning set to true
+   - Set parent_task_id to $CLAUDOMATOR_TASK_ID in each POST body
    - After creating all subtasks, output a brief summary and STOP. Do not implement anything.
+   - You can also specify agent.type (either "claude" or "gemini") to choose the agent for subtasks.
 
 3. If NO — proceed with the task instructions below.