Agent loop state machine refactor, unified LLM interface, and UI improvements

- Rewrite agent loop as Planning→Executing(N)→Completed state machine with
  per-step context isolation to prevent token explosion
- Split tools and prompts by phase (planning vs execution)
- Add advance_step/save_memo tools for step transitions and cross-step memory
- Unify LLM interface: remove duplicate types, single chat_with_tools path
- Add UTF-8 safe truncation (truncate_str) to prevent panics on Chinese text
- Extract CreateForm component, add auto-scroll to execution log
- Add report generation with app access URL, non-blocking title generation
- Add timer system, file serving, app proxy, exec module
- Update Dockerfile with uv, deployment config

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

web/src/ws.ts

@@ -17,12 +17,29 @@ export interface WsWorkflowStatusUpdate {
   status: string
 }
 
+export interface WsRequirementUpdate {
+  type: 'RequirementUpdate'
+  workflow_id: string
+  requirement: string
+}
+
+export interface WsReportReady {
+  type: 'ReportReady'
+  workflow_id: string
+}
+
+export interface WsProjectUpdate {
+  type: 'ProjectUpdate'
+  project_id: string
+  name: string
+}
+
 export interface WsError {
   type: 'Error'
   message: string
 }
 
-export type WsMessage = WsPlanUpdate | WsStepStatusUpdate | WsWorkflowStatusUpdate | WsError
+export type WsMessage = WsPlanUpdate | WsStepStatusUpdate | WsWorkflowStatusUpdate | WsRequirementUpdate | WsReportReady | WsProjectUpdate | WsError
 
 export type WsHandler = (msg: WsMessage) => void