feat: More verbose logging on workflow logs

agent/internal/exec/exec.go

@@ -35,10 +35,21 @@ func (w *streamWriter) Write(p []byte) (int, error) {
 	return len(p), nil
 }
 
+// WorkspacePath returns the per-run working directory for a workspace id. The
+// same id always maps to the same path so RunStep and the cleanup command agree.
+func WorkspacePath(workspaceID string) string {
+	return filepath.Join(os.TempDir(), "vantage-run-"+workspaceID)
+}
+
 // RunStep writes the script to a temp file, provides a WORKFLOW_ENV file for
 // the script to append KEY=value output to, executes it under the requested
 // interpreter, and streams output via emit, returning the terminal result
 // with empty stdout/stderr but populated exit_code/output_env.
+//
+// When the command carries a WorkspaceId the step runs with that per-run working
+// directory as its cwd (created here if missing); the server removes it once the
+// run finishes. The script and env files always live in a private temp dir so
+// they never leak into the shared workspace.
 func RunStep(cmd *pb.RunStepCmd, emit func(seq uint64, data []byte)) *pb.StepResult {
 	res := &pb.StepResult{CommandId: "", OutputEnv: map[string]string{}}
 
@@ -50,6 +61,16 @@ func RunStep(cmd *pb.RunStepCmd, emit func(seq uint64, data []byte)) *pb.StepRes
 	}
 	defer os.RemoveAll(dir)
 
+	workDir := ""
+	if cmd.WorkspaceId != "" {
+		workDir = WorkspacePath(cmd.WorkspaceId)
+		if err := os.MkdirAll(workDir, 0700); err != nil {
+			res.ExitCode = 1
+			res.Stderr = "create workspace: " + err.Error()
+			return res
+		}
+	}
+
 	envFile := filepath.Join(dir, "workflow_env")
 	if err := os.WriteFile(envFile, nil, 0600); err != nil {
 		res.ExitCode = 1
@@ -91,6 +112,10 @@ func RunStep(cmd *pb.RunStepCmd, emit func(seq uint64, data []byte)) *pb.StepRes
 		c = exec.CommandContext(ctx, "bash", scriptPath)
 	}
 
+	if workDir != "" {
+		c.Dir = workDir
+	}
+
 	c.Env = append(os.Environ(), "WORKFLOW_ENV="+envFile)
 	for k, v := range cmd.Env {
 		c.Env = append(c.Env, k+"="+v)

agent/internal/grpc/pb/vantage.pb.go

@@ -63,12 +63,19 @@ type ReportUpdatesResponse struct{}
 type ApplyUpdatesCmd struct{}
 
 type ServerCommand struct {
-	CommandId    string           `json:"command_id"`
-	GenerateKey  *GenerateKeyCmd  `json:"generate_key,omitempty"`
-	DeleteKey    *DeleteKeyCmd    `json:"delete_key,omitempty"`
-	UpdateAgent  *UpdateAgentCmd  `json:"update_agent,omitempty"`
-	ApplyUpdates *ApplyUpdatesCmd `json:"apply_updates,omitempty"`
-	RunStep      *RunStepCmd      `json:"run_step,omitempty"`
+	CommandId        string               `json:"command_id"`
+	GenerateKey      *GenerateKeyCmd      `json:"generate_key,omitempty"`
+	DeleteKey        *DeleteKeyCmd        `json:"delete_key,omitempty"`
+	UpdateAgent      *UpdateAgentCmd      `json:"update_agent,omitempty"`
+	ApplyUpdates     *ApplyUpdatesCmd     `json:"apply_updates,omitempty"`
+	RunStep          *RunStepCmd          `json:"run_step,omitempty"`
+	CleanupWorkspace *CleanupWorkspaceCmd `json:"cleanup_workspace,omitempty"`
+}
+
+// CleanupWorkspaceCmd tells the agent to recursively remove the run's working
+// directory once all steps on that server have finished.
+type CleanupWorkspaceCmd struct {
+	WorkspaceId string `json:"workspace_id"`
 }
 
 type DeleteKeyCmd struct {
@@ -110,6 +117,9 @@ type RunStepCmd struct {
 	Script         string            `json:"script"`
 	Env            map[string]string `json:"env,omitempty"`
 	TimeoutSeconds int               `json:"timeout_seconds,omitempty"`
+	// WorkspaceId names the per-run working directory the agent creates and uses
+	// as the step's cwd. Empty means run in the agent's default directory.
+	WorkspaceId string `json:"workspace_id,omitempty"`
 }
 
 type StepResult struct {

agent/internal/sync/sync.go

@@ -198,6 +198,9 @@ func connectAndHandleStream(ctx context.Context, cfg *config.Config) error {
 		if cmd.ApplyUpdates != nil {
 			go handleApplyUpdates(cfg, cmd)
 		}
+		if cmd.CleanupWorkspace != nil {
+			go handleCleanupWorkspace(cmd)
+		}
 		if cmd.RunStep != nil {
 			go func(rc *pb.RunStepCmd, cid string) {
 				emit := func(seq uint64, data []byte) {
@@ -286,6 +289,16 @@ func handleApplyUpdates(cfg *config.Config, cmd *pb.ServerCommand) {
 	_ = client.ReportUpdates(cfg.ServerID, cfg.AgentToken, nil)
 }
 
+func handleCleanupWorkspace(cmd *pb.ServerCommand) {
+	id := cmd.CleanupWorkspace.WorkspaceId
+	dir := agentexec.WorkspacePath(id)
+	if err := os.RemoveAll(dir); err != nil {
+		log.Printf("cleanup workspace %s failed (cmd=%s): %v", dir, cmd.CommandId, err)
+		return
+	}
+	log.Printf("removed run workspace %s (cmd=%s)", dir, cmd.CommandId)
+}
+
 func handleDeleteKey(cmd *pb.ServerCommand) {
 	label := cmd.DeleteKey.Label
 	keyPath := fmt.Sprintf("/root/.ssh/vantage_%s", strings.ReplaceAll(label, " ", "_"))