Added streaming

orchestration/orchestrator.py

@@ -44,6 +44,13 @@ class ProviderClient:
                          temperature: float, max_tokens: int) -> str:
         return self.call(system, messages, model, temperature, max_tokens)
 
+    async def call_async_streaming(self, system: str, messages: list[dict], model: str,
+                                   temperature: float, max_tokens: int):
+        """Async generator yielding ('thinking'|'content', text) chunks.
+        Default: emits the full response as a single 'content' chunk."""
+        response = await self.call_async(system, messages, model, temperature, max_tokens)
+        yield ('content', response)
+
 
 # --- OpenAI and any OpenAI-compatible endpoint ---
 # Covers AIPA_PROVIDER / {AGENT}_PROVIDER = openai or openai_compatible.
@@ -71,6 +78,29 @@ class OpenAIClient(ProviderClient):
             self.call_async(system, messages, model, temperature, max_tokens)
         )
 
+    async def call_async_streaming(self, system, messages, model, temperature, max_tokens):
+        """Stream via the OpenAI-compatible API, yielding thinking and content chunks.
+        Thinking tokens arrive in delta.reasoning_content (Qwen3, DeepSeek R1 style);
+        regular tokens arrive in delta.content."""
+        full_messages = [{"role": "system", "content": system}] + messages
+        stream = await self.client.chat.completions.create(
+            model=model,
+            temperature=temperature,
+            max_tokens=max_tokens,
+            messages=full_messages,
+            extra_body=self.extra_body or None,
+            stream=True,
+        )
+        async for chunk in stream:
+            if not chunk.choices:
+                continue
+            delta = chunk.choices[0].delta
+            thinking = getattr(delta, 'reasoning_content', None)
+            if thinking:
+                yield ('thinking', thinking)
+            if delta.content:
+                yield ('content', delta.content)
+
 
 # --- Anthropic ---
 # Uncomment when ready to use. Install: pip install anthropic
@@ -283,6 +313,8 @@ async def call_agent_async(
     agent: AgentState,
     user_message: str,
     dry_run: bool = False,
+    stream: bool = True,
+    stream_mode: str = "deliverable",
 ) -> str:
     if dry_run:
         return f"[DRY RUN] Would call {agent.name} ({config.agent_provider_name(agent.name)}) with: {user_message[:100]}..."
@@ -290,6 +322,19 @@ async def call_agent_async(
     agent.add_user(user_message)
 
     try:
+        if stream:
+            response = await ui.stream_agent_output(
+                agent.name,
+                agent.client.call_async_streaming(
+                    system=agent.system_prompt,
+                    messages=agent.messages_for_call(),
+                    model=agent.model,
+                    temperature=agent.temperature,
+                    max_tokens=agent.max_tokens,
+                ),
+                mode=stream_mode,
+            )
+        else:
             response = await agent.client.call_async(
                 system=agent.system_prompt,
                 messages=agent.messages_for_call(),
@@ -302,7 +347,7 @@ async def call_agent_async(
 
     agent.add_assistant(response)
 
-    if config.DEBUG_PRINT_AGENT_OUTPUTS:
+    if config.DEBUG_PRINT_AGENT_OUTPUTS and not stream:
         ui.print_agent_panel(agent.name, response)
 
     return response
@@ -357,7 +402,7 @@ async def dispatch_to_lead(
     if session_id:
         task_store.update_task(task, session_id)
     try:
-        output = await call_agent_async(lead, task.brief, dry_run=dry_run)
+        output = await call_agent_async(lead, task.brief, dry_run=dry_run, stream=False)
         task.output = output
         task.status = "complete"
     except Exception as e:
@@ -424,7 +469,7 @@ async def run_auditor(
         f"{deliverable}"
     )
     auditor.history = []   # Auditor is stateless — fresh call each time
-    return await call_agent_async(auditor, audit_request, dry_run=dry_run)
+    return await call_agent_async(auditor, audit_request, dry_run=dry_run, stream_mode="audit")
 
 
 # ---------------------------------------------------------------------------
@@ -446,7 +491,7 @@ async def update_standing_brief(
         f"log entry at the top of Section 9. Archive anything older than "
         f"{config.SESSION_LOG_RETENTION} sessions."
     )
-    return await call_agent_async(orchestrator, update_request, dry_run=dry_run)
+    return await call_agent_async(orchestrator, update_request, dry_run=dry_run, stream_mode="background")
 
 
 # ---------------------------------------------------------------------------
@@ -502,7 +547,7 @@ class Session:
             f"Directive: {directive}"
         )
         cos_response = await call_agent_async(
-            self.orchestrator, cos_prompt, dry_run=self.dry_run
+            self.orchestrator, cos_prompt, dry_run=self.dry_run, stream=True
         )
 
         # ── Step 2: Parse any task briefs the orchestrator issued ───────────
@@ -532,9 +577,8 @@ class Session:
                 f"Please synthesize these into a final deliverable for the Principal.\n\n"
                 f"{lead_outputs_text}"
             )
-            with ui.synthesis_progress():
             deliverable = await call_agent_async(
-                    self.orchestrator, synthesis_prompt, dry_run=self.dry_run
+                self.orchestrator, synthesis_prompt, dry_run=self.dry_run, stream=True
             )
 
             dispatched_leads = sorted({t.assigned_to for t in completed_tasks})
@@ -549,7 +593,6 @@ class Session:
         audit_memo = ""
         if self.auto_audit:
             try:
-                with ui.audit_progress():
                 audit_memo = await run_auditor(
                     self.auditor, deliverable, task_id, dry_run=self.dry_run
                 )
@@ -561,7 +604,6 @@ class Session:
 
     async def close(self):
         summary = "\n".join(self.session_log)
-        with ui.brief_update_progress():
         updated_brief = await update_standing_brief(
             self.orchestrator, summary, dry_run=self.dry_run
         )
@@ -645,13 +687,22 @@ async def cmd_history(session: Session, args: str) -> bool:
     return True
 
 
+@command(["/thinking"], "Show the thinking output from the last response")
+async def cmd_thinking(session: Session, args: str) -> bool:
+    last = ui.get_last_thinking()
+    if not last:
+        ui.print_system("No thinking output recorded — thinking mode may not be active.")
+    else:
+        ui.print_thinking_expansion(last)
+    return True
+
+
 @command(["/audit"], "Re-run Vera's audit on the last deliverable")
 async def cmd_audit(session: Session, args: str) -> bool:
     if not session.last_deliverable:
         ui.print_warning("No deliverable yet — issue a directive first.")
         return True
     ui.print_system("Running audit on last deliverable…")
-    with ui.audit_progress():
     memo = await run_auditor(
         session.auditor, session.last_deliverable,
         session.last_task_id, dry_run=session.dry_run,
@@ -731,7 +782,15 @@ async def direct_agent_session(agent_name: str, dry_run: bool = False):
         if not user_input or user_input.lower() in ("/quit", "/exit", "/q"):
             break
 
-        response = await call_agent_async(agent, user_input, dry_run=dry_run)
+        if user_input.lower() == "/thinking":
+            last = ui.get_last_thinking()
+            if not last:
+                ui.print_system("No thinking output recorded — thinking mode may not be active.")
+            else:
+                ui.print_thinking_expansion(last)
+            continue
+
+        response = await call_agent_async(agent, user_input, dry_run=dry_run, stream=True)
         ui.print_agent_panel(agent_name, response)
 
     ui.print_session_footer(0)

orchestration/ui.py

@@ -23,8 +23,9 @@ from datetime import datetime
 
 import config as _config
 
-from rich.console import Console
+from rich.console import Console, Group
 from rich.columns import Columns
+from rich.live import Live
 from rich.markdown import Markdown
 from rich.panel import Panel
 from rich.progress import (
@@ -75,6 +76,10 @@ def _build_theme() -> Theme:
 
 console = Console(theme=_build_theme(), highlight=False)
 
+_last_thinking: str = ""    # Stored from most recent streamed response
+_last_stream_rendered: bool = False  # True when stream_agent_output rendered the final panel
+
+
 # ---------------------------------------------------------------------------
 # Agent visual config
 # ---------------------------------------------------------------------------
@@ -268,7 +273,13 @@ def print_agent_panel(agent_name: str, content: str, task_id: str = ""):
     """
     Render an agent's output in a styled panel.
     Content is treated as Markdown.
+    Skips printing if stream_agent_output already rendered this response.
     """
+    global _last_stream_rendered
+    if _last_stream_rendered:
+        _last_stream_rendered = False
+        return
+
     color      = _agent_color(agent_name)
     title_text = _agent_display_title(agent_name)
     subtitle   = f"[dim]{task_id}[/dim]" if task_id else ""
@@ -286,15 +297,51 @@ def print_agent_panel(agent_name: str, content: str, task_id: str = ""):
 
 
 def print_deliverable(content: str, task_id: str = ""):
-    """Orchestrator's synthesized deliverable."""
+    """
-    print_agent_panel(_config.ORCHESTRATOR_AGENT, content, task_id)
+    Orchestrator's synthesized deliverable.
+    Skips printing if stream_agent_output already rendered this response.
+    """
+    global _last_stream_rendered
+    if _last_stream_rendered:
+        _last_stream_rendered = False
+        return
+    _print_deliverable_panel(content, task_id)
+
+
+def _print_deliverable_panel(content: str, task_id: str = ""):
+    """Internal: always render the deliverable panel unconditionally."""
+    agent_name = _config.ORCHESTRATOR_AGENT
+    color      = _agent_color(agent_name)
+    title_text = _agent_display_title(agent_name)
+    subtitle   = f"[dim]{task_id}[/dim]" if task_id else ""
+
+    panel = Panel(
+        Markdown(content),
+        title=f"[{color}] {title_text} [/{color}]",
+        subtitle=subtitle,
+        border_style=color,
+        padding=(1, 2),
+        box=box.ROUNDED,
+    )
+    console.print()
+    console.print(panel)
 
 
 def print_audit_memo(content: str, task_id: str = ""):
     """
     Auditor's memo. Verdict line drives the border colour.
     PASS = green, PASS WITH NOTES = yellow, FLAG = yellow, REJECT = red.
+    Skips printing if stream_agent_output already rendered this response.
     """
+    global _last_stream_rendered
+    if _last_stream_rendered:
+        _last_stream_rendered = False
+        return
+    _print_audit_memo_panel(content, task_id)
+
+
+def _print_audit_memo_panel(content: str, task_id: str = ""):
+    """Internal: always render the audit memo panel with verdict-based border."""
     aud_name  = _config.AUDITOR_AGENT
     aud_title = _config.agent_title(aud_name)
     aud_color = _agent_color(aud_name)
@@ -390,6 +437,267 @@ def brief_update_progress():
         progress.add_task("", total=None)
         yield
 
+# ---------------------------------------------------------------------------
+# Streaming output with thinking support
+# ---------------------------------------------------------------------------
+
+def _thinking_panel_title(agent_name: str, mode: str) -> str:
+    """Return the thinking panel title, tinted by agent identity."""
+    if mode == "audit":
+        color = _agent_color(agent_name)
+        return f"[{color}]Vera thinking…[/{color}]"
+    color = _agent_color(agent_name)
+    return f"[{color} dim]Thinking…[/{color} dim]"
+
+
+def _build_thinking_renderable(
+    thinking_text: str,
+    active: bool,
+    agent_name: str,
+    mode: str,
+    window: int = 300,
+) -> object:
+    """
+    Build the thinking panel (active) or collapsed summary line (done).
+
+    Active:   dim panel with a rolling window of the last `window` chars,
+              plus an overflow indicator when earlier content has scrolled off.
+    Collapsed: single dim line — '  ▸ Thinking (N chars, L lines)'.
+    """
+    if not thinking_text:
+        return None
+
+    if active:
+        overflow = len(thinking_text) - window
+        if overflow > 0:
+            preview = thinking_text[-window:]
+            header = Text(f"  +{overflow} chars earlier\n", style="dim italic")
+            body   = Text(preview, style="dim")
+            content = Group(header, body)
+        else:
+            content = Text(thinking_text, style="dim")
+
+        return Panel(
+            content,
+            title=_thinking_panel_title(agent_name, mode),
+            border_style="dim",
+            padding=(0, 1),
+            box=box.SIMPLE,
+        )
+    else:
+        lines = thinking_text.count("\n") + 1
+        return Text(
+            f"  ▸ Thinking  [{len(thinking_text):,} chars · {lines} lines]",
+            style="dim",
+        )
+
+
+async def stream_agent_output(
+    agent_name: str,
+    stream_gen,
+    mode: str = "deliverable",
+) -> str:
+    """
+    Consume an async generator of ('thinking'|'content', text) chunks and
+    render them live in the terminal.
+
+    mode values
+    -----------
+    "deliverable"  Normal agent response — thinking panel + full styled panel.
+                   The Live display is NOT transient: the final Markdown panel
+                   is rendered inside the Live block, so there is no erase/
+                   reprint flash. Sets _last_stream_rendered = True so that
+                   the caller's print_deliverable / print_agent_panel becomes
+                   a no-op.
+
+    "audit"        Like "deliverable" but uses Vera's yellow identity cues
+                   during the thinking phase.  Sets _last_stream_rendered so
+                   print_audit_memo is a no-op (the verdict-border panel is
+                   rendered here instead).
+
+    "background"   Housekeeping call (e.g. standing-brief update).  A compact
+                   transient progress line is shown; no full panel is printed.
+                   _last_stream_rendered is set so the caller's print_* is a
+                   no-op.
+
+    In all modes the full content string is returned and _last_thinking is
+    updated so /thinking works.
+
+    Thinking tokens stream inside a dim panel.  When the first content token
+    arrives the panel collapses to a single dim summary line and content begins
+    streaming in the agent's normal styled panel.
+
+    The full thinking text is stored in _last_thinking and can be retrieved
+    with get_last_thinking() (exposed to the /thinking command).
+
+    Returns the full content string.
+    """
+    global _last_thinking, _last_stream_rendered
+
+    color      = _agent_color(agent_name)
+    title_text = _agent_display_title(agent_name)
+
+    # ── Background mode: compact transient progress line ────────────────────
+    if mode == "background":
+        thinking_parts: list[str] = []
+        content_parts:  list[str] = []
+
+        with Progress(
+            SpinnerColumn(style="dim"),
+            TextColumn("[dim]Updating standing brief…[/dim]"),
+            TimeElapsedColumn(),
+            console=console,
+            transient=True,
+        ) as prog:
+            prog.add_task("", total=None)
+            async for chunk_type, text in stream_gen:
+                if chunk_type == "thinking":
+                    thinking_parts.append(text)
+                elif chunk_type == "content":
+                    content_parts.append(text)
+
+        _last_thinking       = "".join(thinking_parts)
+        _last_stream_rendered = True
+        return "".join(content_parts)
+
+    # ── Deliverable / audit modes ────────────────────────────────────────────
+    thinking_parts: list[str] = []
+    content_parts:  list[str] = []
+    phase = "start"  # "start" | "thinking" | "content"
+
+    def _build_final_panel(content_text: str) -> Panel:
+        """Render the agent's finished output as a polished Markdown panel."""
+        if mode == "audit":
+            # Re-derive verdict border from completed content
+            aud_name  = _config.AUDITOR_AGENT
+            aud_title = _config.agent_title(aud_name)
+            aud_color = _agent_color(aud_name)
+            verdict_style = "dim"
+            for line in content_text.splitlines():
+                upper = line.upper()
+                if "VERDICT:" in upper:
+                    if "REJECT" in upper:
+                        verdict_style = "reject"
+                    elif "FLAG" in upper:
+                        verdict_style = "flag"
+                    elif "PASS WITH NOTES" in upper:
+                        verdict_style = "warning"
+                    elif "PASS" in upper:
+                        verdict_style = "pass"
+                    break
+            return Panel(
+                Markdown(content_text),
+                title=f"[{aud_color}] {aud_name.capitalize()} — {aud_title} [/{aud_color}]",
+                border_style=verdict_style if verdict_style != "dim" else aud_color,
+                padding=(1, 2),
+                box=box.ROUNDED,
+            )
+        else:
+            return Panel(
+                Markdown(content_text),
+                title=f"[{color}] {title_text} [/{color}]",
+                border_style=color,
+                padding=(1, 2),
+                box=box.ROUNDED,
+            )
+
+    def _renderable(final: bool = False):
+        """
+        Build the live renderable for the current stream state.
+
+        final=True: replace plain Text content with Markdown and apply the
+        verdict-aware border so the Live display ends in its final polished
+        form.  This avoids a transient erase + reprint flash.
+        """
+        thinking_text = "".join(thinking_parts)
+        content_text  = "".join(content_parts)
+        items: list   = []
+
+        # ── Agent identity line (shown while only thinking, not yet content) ──
+        if phase in ("start", "thinking") and not content_text:
+            items.append(Text(
+                f"  {title_text}",
+                style=f"dim {color}",
+            ))
+
+        # ── Thinking panel / collapsed summary ──────────────────────────────
+        thinking_active = (phase == "thinking")
+        think_renderable = _build_thinking_renderable(
+            thinking_text, thinking_active, agent_name, mode
+        )
+        if think_renderable is not None:
+            items.append(think_renderable)
+
+        # ── Content panel ────────────────────────────────────────────────────
+        if content_text or phase == "content":
+            if final and content_text:
+                items.append(_build_final_panel(content_text))
+            else:
+                items.append(Panel(
+                    Text(content_text),
+                    title=f"[{color}] {title_text} [/{color}]",
+                    border_style=color,
+                    padding=(1, 2),
+                    box=box.ROUNDED,
+                ))
+
+        if not items:
+            # Nothing yet — show a faint connecting indicator
+            return Text(
+                f"  [{_agent_color(agent_name)} dim]{title_text}  …[/{_agent_color(agent_name)} dim]",
+                style="dim",
+            )
+        return Group(*items) if len(items) > 1 else items[0]
+
+    console.print()
+    # transient=False: the Live display persists — we render the final panel
+    # inside it before stopping, so there is no erase/reprint cycle.
+    with Live(
+        _renderable(),
+        console=console,
+        refresh_per_second=15,
+        transient=False,
+        vertical_overflow="visible",
+    ) as live:
+        async for chunk_type, text in stream_gen:
+            if chunk_type == "thinking":
+                if phase == "start":
+                    phase = "thinking"
+                thinking_parts.append(text)
+            elif chunk_type == "content":
+                if phase in ("start", "thinking"):
+                    phase = "content"
+                content_parts.append(text)
+            live.update(_renderable())
+
+        # Final update: replace plain Text with formatted Markdown panel
+        # (and apply verdict border for audit mode) before the Live context
+        # exits and the display freezes in place.
+        live.update(_renderable(final=True))
+
+    _last_thinking        = "".join(thinking_parts)
+    _last_stream_rendered = True
+    return "".join(content_parts)
+
+
+def get_last_thinking() -> str:
+    """Return the thinking text captured from the most recent streamed response."""
+    return _last_thinking
+
+
+def print_thinking_expansion(thinking_text: str):
+    """Render the stored thinking output in a dim panel."""
+    panel = Panel(
+        Text(thinking_text, style="dim"),
+        title="[dim] Thinking — last response [/dim]",
+        border_style="dim",
+        padding=(1, 2),
+        box=box.ROUNDED,
+    )
+    console.print()
+    console.print(panel)
+
+
 # ---------------------------------------------------------------------------
 # Dry-run callout
 # ---------------------------------------------------------------------------