zwarm 3.10.2__py3-none-any.whl → 3.10.5__py3-none-any.whl

zwarm/tools/delegation.py

@@ -34,17 +34,43 @@ if TYPE_CHECKING:
 ADAPTERS = ["codex", "claude"]
 
 
-def _get_session_manager(orchestrator: "Orchestrator"):
+def _get_all_adapter_managers(orchestrator: "Orchestrator") -> dict:
     """
-    Get the default session manager for list/get operations.
+    Get session managers for ALL adapters.
 
-    Uses CodexSessionManager as the default since all adapters share
-    the same .zwarm/sessions/ directory structure.
+    This ensures we can find sessions regardless of which adapter created them.
+    Returns a dict of {adapter_name: manager}.
     """
-    if not hasattr(orchestrator, "_session_manager") or orchestrator._session_manager is None:
-        from zwarm.sessions import CodexSessionManager
-        orchestrator._session_manager = CodexSessionManager(orchestrator.working_dir / ".zwarm")
-    return orchestrator._session_manager
+    # Initialize adapter managers dict if needed
+    if not hasattr(orchestrator, "_adapter_managers"):
+        orchestrator._adapter_managers = {}
+
+    # Ensure all adapters have managers
+    from zwarm.sessions import get_session_manager
+    for adapter in ADAPTERS:
+        if adapter not in orchestrator._adapter_managers:
+            manager = get_session_manager(adapter, str(orchestrator.working_dir / ".zwarm"))
+            orchestrator._adapter_managers[adapter] = manager
+
+    return orchestrator._adapter_managers
+
+
+def _get_session_with_manager(orchestrator: "Orchestrator", session_id: str):
+    """
+    Find a session across all adapters and return (session, manager).
+
+    Since sessions are adapter-scoped, we need to check each adapter's
+    manager to find where the session lives.
+
+    Returns:
+        (session, manager) if found, (None, None) if not found
+    """
+    managers = _get_all_adapter_managers(orchestrator)
+    for adapter, manager in managers.items():
+        session = manager.get_session(session_id)
+        if session:
+            return session, manager
+    return None, None
 
 
 def _get_adapter_manager(orchestrator: "Orchestrator", adapter: str):
@@ -177,7 +203,7 @@ def delegate(
     WORKFLOW:
         1. delegate(task="...", model="5.2") -> session_id
         2. sleep(30)
-        3. peek_session(session_id) -> {is_running: true/false}
+        3. list_sessions() -> check which are done (needs_attention=True)
         4. If is_running, goto 2
         5. check_session(session_id) -> FULL response
 
@@ -295,9 +321,8 @@ def converse(
         sleep(30)
         check_session(session_id)  # Get response
     """
-    # First get session to determine adapter
-    default_manager = _get_session_manager(self)
-    session = default_manager.get_session(session_id)
+    # Find session across all adapters
+    session, manager = _get_session_with_manager(self, session_id)
 
     if not session:
         return {
@@ -322,9 +347,8 @@ def converse(
             "hint": "Start a new session with delegate()",
         }
 
-    # Get the correct adapter manager for this session
+    # Use the manager that found the session (correct adapter)
     adapter = getattr(session, "adapter", "codex")
-    manager = _get_adapter_manager(self, adapter)
 
     # Inject the follow-up message
     # This uses the adapter's inject_message() which:
@@ -356,25 +380,22 @@ def converse(
 def check_session(
     self: "Orchestrator",
     session_id: str,
+    latest: bool = True,
 ) -> dict[str, Any]:
     """
-    Check the status of a session and get the FULL response.
+    Check the status of a session and get the response.
 
     This is your primary tool for seeing what an executor accomplished.
-    Returns the complete, untruncated response from the agent.
-
-    Use this after peek_session() shows the session is done, or when
-    you need to see the full details of what was accomplished.
 
     Args:
         session_id: The session to check.
+        latest: If True (default), only return the latest assistant message.
+                If False, return full conversation history.
 
     Returns:
-        {session_id, status, response (FULL), tokens, runtime}
+        {session_id, status, response, tokens, runtime}
     """
-    manager = _get_session_manager(self)
-
-    session = manager.get_session(session_id)
+    session, manager = _get_session_with_manager(self, session_id)
     if not session:
         return {
             "success": False,
@@ -382,13 +403,22 @@ def check_session(
             "hint": "Use list_sessions() to see available sessions",
         }
 
-    # Get latest response - FULL, not truncated
-    response_text = ""
     messages = manager.get_messages(session_id)
-    for msg in reversed(messages):
-        if msg.role == "assistant":
-            response_text = msg.content  # Full content, no truncation
-            break
+
+    if latest:
+        # Only get the latest assistant message
+        response_text = ""
+        for msg in reversed(messages):
+            if msg.role == "assistant":
+                response_text = msg.content
+                break
+        response_content = response_text if response_text else "(no response yet)"
+    else:
+        # Return full conversation history
+        response_content = [
+            {"role": msg.role, "content": msg.content}
+            for msg in messages
+        ]
 
     # Build log path
     log_path = str(manager._output_path(session.id, session.turn))
@@ -401,8 +431,8 @@ def check_session(
         "is_running": session.is_running,
         "turn": session.turn,
         "message_count": len(messages),
-        "task": _truncate(session.task, 80),  # Task can stay truncated
-        "response": response_text if response_text else "(no response yet)",  # FULL response
+        "task": _truncate(session.task, 80),
+        "response": response_content,
         "tokens": _get_total_tokens(session),
         "runtime": session.runtime,
         "log_file": log_path,
@@ -417,51 +447,6 @@ def check_session(
     return result
 
 
-@weaveTool
-def peek_session(
-    self: "Orchestrator",
-    session_id: str,
-) -> dict[str, Any]:
-    """
-    Quick peek at a session - minimal info for FAST POLLING.
-
-    Use this in your polling loop to check if a session is done:
-        1. delegate() -> start work
-        2. sleep(30)
-        3. peek_session() -> is_running? If yes, goto 2
-        4. check_session() -> get FULL response
-
-    Returns truncated preview only. Once done, use check_session() for full response.
-
-    Args:
-        session_id: The session to peek at.
-
-    Returns:
-        {session_id, status, is_running, latest_message (truncated preview)}
-    """
-    manager = _get_session_manager(self)
-
-    session = manager.get_session(session_id)
-    if not session:
-        return {"success": False, "error": f"Unknown session: {session_id}"}
-
-    # Get latest assistant message only
-    latest = ""
-    messages = manager.get_messages(session_id)
-    for msg in reversed(messages):
-        if msg.role == "assistant":
-            latest = msg.content.replace("\n", " ")
-            break
-
-    return {
-        "success": True,
-        "session_id": session.short_id,
-        "status": session.status.value,
-        "is_running": session.status.value == "running",
-        "latest_message": _truncate(latest, 150) if latest else None,
-    }
-
-
 @weaveTool
 def get_trajectory(
     self: "Orchestrator",
@@ -487,9 +472,7 @@ def get_trajectory(
     - check_session() -> what did the agent conclude? (FULL response)
     - get_trajectory() -> what steps did the agent take? (step-by-step)
     """
-    manager = _get_session_manager(self)
-
-    session = manager.get_session(session_id)
+    session, manager = _get_session_with_manager(self, session_id)
     if not session:
         return {"success": False, "error": f"Unknown session: {session_id}"}
 
@@ -565,9 +548,7 @@ def end_session(
     Returns:
         {session_id, status}
     """
-    manager = _get_session_manager(self)
-
-    session = manager.get_session(session_id)
+    session, manager = _get_session_with_manager(self, session_id)
     if not session:
         return {
             "success": False,
@@ -594,7 +575,7 @@ def end_session(
                 "session_id": session_id,
             }
 
-        # Refresh
+        # Refresh - use same manager
         session = manager.get_session(session_id)
 
     return {
@@ -633,7 +614,8 @@ def list_sessions(
     """
     from datetime import datetime
 
-    manager = _get_session_manager(self)
+    # Get managers for ALL adapters to aggregate sessions
+    managers = _get_all_adapter_managers(self)
 
     # Map string status to enum
     from zwarm.sessions import SessionStatus
@@ -648,7 +630,13 @@ def list_sessions(
         }
         status_filter = status_map.get(status.lower())
 
-    sessions = manager.list_sessions(status=status_filter)
+    # Aggregate sessions from ALL adapters
+    sessions = []
+    for manager in managers.values():
+        sessions.extend(manager.list_sessions(status=status_filter))
+
+    # Sort by created_at descending (newest first)
+    sessions.sort(key=lambda s: s.created_at, reverse=True)
 
     def time_ago(iso_str: str) -> tuple[str, float]:
         """Convert ISO timestamp to ('Xm ago', seconds)."""
@@ -681,13 +669,16 @@ def list_sessions(
 
         updated_str, updated_secs = time_ago(s.updated_at)
 
-        # Get last assistant message
-        messages = manager.get_messages(s.id)
+        # Get last assistant message using the correct adapter's manager
+        session_adapter = getattr(s, "adapter", "codex")
+        session_manager = managers.get(session_adapter)
         last_message = ""
-        for msg in reversed(messages):
-            if msg.role == "assistant":
-                last_message = msg.content.replace("\n", " ")
-                break
+        if session_manager:
+            messages = session_manager.get_messages(s.id)
+            for msg in reversed(messages):
+                if msg.role == "assistant":
+                    last_message = msg.content.replace("\n", " ")
+                    break
 
         # Flag sessions that need attention:
         # - Recently completed (< 60s)
@@ -739,10 +730,11 @@ def sleep(self, seconds: float) -> dict[str, Any]:
     give them time to complete before checking their status. This lets you
     manage your own polling loop:
 
-    1. delegate(task, wait=False) -> start background work
-    2. sleep(10) -> wait a bit
-    3. peek_session(id) -> check if done
+    1. delegate(task) -> start background work
+    2. sleep(30) -> wait a bit
+    3. list_sessions() -> check which are done (needs_attention=True)
     4. Repeat 2-3 if still running
+    5. check_session(id) -> get result
 
     Args:
         seconds: Number of seconds to sleep (max 300 = 5 minutes)

zwarm/watchers/llm_watcher.py

@@ -58,7 +58,7 @@ def _extract_tool_call_summary(tc: Any) -> str:
         elif name == "bash":
             cmd = args.get("command", "")[:60]
             return f"$ {cmd}"
-        elif name in ("check_session", "peek_session", "end_session"):
+        elif name in ("check_session", "end_session"):
             sid = args.get("session_id", "")[:8]
             return f"{name}({sid})"
         elif name == "list_sessions":

{zwarm-3.10.2.dist-info → zwarm-3.10.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: zwarm
-Version: 3.10.2
+Version: 3.10.5
 Summary: Multi-Agent CLI Orchestration Research Platform
 Requires-Python: <3.14,>=3.13
 Requires-Dist: prompt-toolkit>=3.0.52
@@ -87,14 +87,18 @@ Want a 3-minute walkthrough? See `docs/DEMO.md` for a pilot + interactive demo.
 
 ## Multi-Adapter Support
 
-zwarm supports multiple executor backends:
+zwarm supports multiple executor backends with simple model shortcuts:
 
-| Adapter | CLI | Models | Config |
-|---------|-----|--------|--------|
-| **Codex** | `codex` | gpt-5.1-codex-mini, etc. | `.zwarm/codex.toml` |
-| **Claude** | `claude` | sonnet, opus, haiku | `.zwarm/claude.toml` |
+| Model | Alias | Description |
+|-------|-------|-------------|
+| `gpt-5.2-codex` | `5.2` | GPT-5.2 Codex - fast, great for code (default) |
+| `gpt-5.2` | `5.2-think` | GPT-5.2 with extended reasoning |
+| `sonnet` | - | Claude Sonnet - balanced |
+| `opus` | - | Claude Opus - most capable |
 
-You can mix adapters in the same session - for example, use Claude Opus for complex reasoning tasks and Codex Mini for quick edits.
+**Adapter is auto-detected from model name** - just use `model="opus"` and zwarm handles the rest.
+
+Mix models freely - use Opus for complex reasoning, 5.2 for quick edits.
 
 ---
 
@@ -184,7 +188,7 @@ zwarm interactive
 
 | Command | Description |
 |---------|-------------|
-| `spawn "task" [--search]` | Start a new session (--search enables web) |
+| `spawn "task" [--model M]` | Start a new session (model: 5.2, opus, sonnet) |
 | `ls` | Dashboard of all sessions (with costs, models) |
 | `? ID` / `peek ID` | Quick status check |
 | `show ID` | Full session details |
@@ -213,8 +217,8 @@ $ zwarm interactive
 ⟳ 2 running
 
 ID       │    │ Task                          │ Model        │ Tokens  │ Cost
-abc123   │ ⟳  │ Add tests for the auth...     │ codex-mini   │ 5,234   │ $0.052
-def456   │ ⟳  │ Fix type errors in utils...   │ codex-mini   │ 2,100   │ $0.021
+abc123   │ ⟳  │ Add tests for the auth...     │ 5.2-codex   │ 5,234   │ $0.052
+def456   │ ⟳  │ Fix type errors in utils...   │ 5.2-codex   │ 2,100   │ $0.021
 
 > watch abc123
 Watching abc123... (Ctrl+C to stop)
@@ -254,17 +258,20 @@ The orchestrator LLM has access to:
 
 | Tool | Description |
 |------|-------------|
-| `delegate(task, adapter="codex")` | Start a new coding session |
+| `delegate(task, model="5.2")` | Start a new coding session |
 | `converse(id, msg)` | Continue a session |
 | `check_session(id)` | Get full session details |
 | `peek_session(id)` | Quick status check |
+| `get_trajectory(id)` | See what steps the agent took |
 | `list_sessions()` | List all sessions |
 | `end_session(id)` | Kill/delete a session |
 | `sleep(seconds)` | Wait before checking again |
+| `bash(cmd)` | Run verification commands (tests, linters) |
+| `exit()` | Signal task completion |
 
 **Async-first**: All sessions run in the background. The orchestrator uses `sleep()` to wait, then checks on progress.
 
-**Multi-adapter**: Pass `adapter="claude"` or `adapter="codex"` to `delegate()` to choose the backend.
+**Model shortcuts**: Just use `model="5.2"` or `model="opus"` - the adapter is auto-detected.
 
 **Web Search**: Enable `web_search=True` in config for tasks needing current info (API docs, latest releases, etc.).
 
@@ -361,14 +368,14 @@ enabled = ["progress", "budget", "delegation", "delegation_reminder"]
 
 **`.zwarm/codex.toml`** - Controls the Codex CLI:
 ```toml
-model = "gpt-5.1-codex-mini"
+model = "gpt-5.2-codex"  # or gpt-5.2 for extended reasoning
 model_reasoning_effort = "high"  # low | medium | high
-full_auto = true
+full_danger = true  # Skip approval prompts
 ```
 
 **`.zwarm/claude.toml`** - Controls the Claude Code CLI:
 ```toml
-model = "sonnet"  # sonnet | opus | haiku
+model = "opus"  # opus | sonnet
 full_danger = true  # Skip permission prompts
 ```
 

{zwarm-3.10.2.dist-info → zwarm-3.10.5.dist-info}/RECORD

@@ -1,39 +1,42 @@
 zwarm/__init__.py,sha256=3i3LMjHwIzE-LFIS2aUrwv3EZmpkvVMe-xj1h97rcSM,837
-zwarm/orchestrator.py,sha256=CqZpe4h8-Z7rpy-boN2W9-sZdFfLONOnNMp2MJP2pbc,25292
+zwarm/orchestrator.py,sha256=1nQmFTUKzrtvrKi08gg9ow6UHt-KvNozXz1td-yrCbA,26878
 zwarm/test_orchestrator_watchers.py,sha256=QpoaehPU7ekT4XshbTOWnJ2H0wRveV3QOZjxbgyJJLY,807
 zwarm/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-zwarm/cli/interactive.py,sha256=81aeiZIrPv5XquEkgDPkl_yYUYSGGhjvtcIVFV7jK_c,41716
-zwarm/cli/main.py,sha256=xQb004Bdm5ghOGqnBvlUIceuR4A8SSSxF3wJcx_aFZU,77369
-zwarm/cli/pilot.py,sha256=ABqPftf3maXhVrvmklpgZuPUELhseM0Tvjdn2_DANXM,42532
+zwarm/cli/interactive.py,sha256=Bd2lBKV1qnVjEx3-swCBl0U0T7JlFaCX9JmSxhffr84,41704
+zwarm/cli/main.py,sha256=WlEpdsdGacoJcSOk3_Rhr9athUGvYcDjixsrFsXg8fE,77251
+zwarm/cli/pilot.py,sha256=zq7p-R4dflY2UocbDNdDOpok4Lqu97CmzTTxM5Ji1Ow,42331
+zwarm/compression/__init__.py,sha256=tMx9jBVwrxKBtim8l4DdKllsE77yoq0kcYn7KG7tFAU,936
+zwarm/compression/rollout_compression.py,sha256=Y5xjIkLDaCdTilzagbjpS7cpCoAuKpoyG9TC5jR4Hu8,9647
+zwarm/compression/tc_compression.py,sha256=_qBKOWcCmqsU3EIg8_rJzeEgUL9lo4JcOnl2yPleTKQ,4867
 zwarm/core/__init__.py,sha256=nEdpEHMFo0gEEKgX-eKHabyOdrOI6UXfWqLu3FfZDao,376
 zwarm/core/checkpoints.py,sha256=D6sXCMB7Sa1kchQ9_lQx_rabwc5-_7jbuynWgA1nkNY,6560
 zwarm/core/compact.py,sha256=Y8C7Gs-5-WOU43WRvQ863Qzd5xtuEqR6Aw3r2p8_-i8,10907
-zwarm/core/config.py,sha256=m3Vm6U_BNtEDu_cz2d6E3p_RNQfRHWaq-946mDru9-8,12656
+zwarm/core/config.py,sha256=jjn-BV5CME_PgIB8gW3o20b5qSNUdPoWWMZp22jnVmg,13813
 zwarm/core/costs.py,sha256=Z-5o-ZQWRCfFv0mTHev4Ke1AzyXKhXWO6ss7S8eBX9U,1485
 zwarm/core/environment.py,sha256=v7wwVCTIOt_qfiJEe774oM4vIYnlb28s6LJXucJdjoo,8735
 zwarm/core/models.py,sha256=PrC3okRBVJxISUa1Fax4KkagqLT6Xub-kTxC9drN0sY,10083
-zwarm/core/registry.py,sha256=MwVWpn9HWC8jwwkv94PFFbul-oY-vO0sTUKocGUNVx8,9551
+zwarm/core/registry.py,sha256=ZRN2vCUU-XSVnAibGefHOb9B5hEPc-i0uh-Vr04bsHU,8935
 zwarm/core/state.py,sha256=MzrvODKEiJovI7YI1jajW4uukineZ3ezmW5oQinMgjg,11563
 zwarm/core/test_compact.py,sha256=WSdjCB5t4YMcknsrkmJIUsVOPY28s4y9GnDmu3Z4BFw,11878
 zwarm/core/test_config.py,sha256=bXXd3OHhK-ndC7wAxePWIdpu73s4O1eScxi3xDzrZwA,4828
 zwarm/core/test_models.py,sha256=sWTIhMZvuLP5AooGR6y8OR2EyWydqVfhmGrE7NPBBnk,8450
 zwarm/prompts/__init__.py,sha256=DI307o712F8qQyDt5vwnFgpVBrxpKwjhr0MaBHLzr9E,334
-zwarm/prompts/orchestrator.py,sha256=rfxpVCNAlTdQN8h0hgtU1OOr_9Io62ULZFisBjNUZVs,15076
-zwarm/prompts/pilot.py,sha256=LHo9JwOmCXcPvxXBqztYFVBt6djUhenTIW-QjppQ1vg,6041
+zwarm/prompts/orchestrator.py,sha256=HPoJxtH4_4yLHv4sS2SCYRsUjQo5S7_uGCodm-K97gA,6979
+zwarm/prompts/pilot.py,sha256=ZKrxJHIni3cnT-oWQmRV-S-X_PkdiJnEiLGgCjism3E,6412
 zwarm/sessions/__init__.py,sha256=5fPkl6JRS_GwPn9hi5iv3dzIpGWu_yghPtvPZdujhnM,1728
 zwarm/sessions/base.py,sha256=3YBd-WWKslQvsBtu03Blth8cEGc_4k4H3GOoKJoTcgg,16976
 zwarm/sessions/claude.py,sha256=hBP_TpNFJjR29IRGJFB3rlG7Z9uWEYSbBGV61tpIr00,16672
-zwarm/sessions/manager.py,sha256=Vq5PePzKfy658EVG24SFsUMXQc1OGgOm8vdOX_WPMF8,18530
+zwarm/sessions/manager.py,sha256=g_QQM9sGdpQ1MK1jdwWMrADeJZY2AqfGDBLVtQasUxg,18520
 zwarm/tools/__init__.py,sha256=FpqxwXJA6-fQ7C-oLj30jjK_0qqcE7MbI0dQuaB56kU,290
-zwarm/tools/delegation.py,sha256=NFMX-f05r28A1OgzYaSMdrq_8VPpP1pJ_nfouYr_2zA,24690
+zwarm/tools/delegation.py,sha256=UTDueeLIpYWMbCN1klPWBUno1zvkhPwqfzO4OtTQiNw,24765
 zwarm/watchers/__init__.py,sha256=a96s7X6ruYkF2ItWWOZ3Q5QUOMOoeCW4Vz8XXcYLXPM,956
 zwarm/watchers/base.py,sha256=r1GoPlj06nOT2xp4fghfSjxbRyFFFQUB6HpZbEyO2OY,3834
 zwarm/watchers/builtin.py,sha256=IL5QwwKOIqWEfJ_uQWb321Px4i5OLtI_vnWQMudqKoA,19064
-zwarm/watchers/llm_watcher.py,sha256=yJGpE3BGKNZX3qgPsiNtJ5d3UJpiTT1V-A-Rh4AiMYM,11029
+zwarm/watchers/llm_watcher.py,sha256=bj7snjMdSbarZA9YSPHa660pTU3TnJ6RQ8rSf6YX8ys,11013
 zwarm/watchers/manager.py,sha256=XZjBVeHjgCUlkTUeHqdvBvHoBC862U1ik0fG6nlRGog,5587
 zwarm/watchers/registry.py,sha256=A9iBIVIFNtO7KPX0kLpUaP8dAK7ozqWLA44ocJGnOw4,1219
 zwarm/watchers/test_watchers.py,sha256=zOsxumBqKfR5ZVGxrNlxz6KcWjkcdp0QhW9WB0_20zM,7855
-zwarm-3.10.2.dist-info/METADATA,sha256=GPtjlVc851Hj_7-81_svlrLhdMz0ZpHptPrMjFnkrrs,11422
-zwarm-3.10.2.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-zwarm-3.10.2.dist-info/entry_points.txt,sha256=u0OXq4q8d3yJ3EkUXwZfkS-Y8Lcy0F8cWrcQfoRxM6Q,46
-zwarm-3.10.2.dist-info/RECORD,,
+zwarm-3.10.5.dist-info/METADATA,sha256=XiRWnNB4N5oHNFaweJbjoDT4-5BHiNSdcBniJTk1lFE,11761
+zwarm-3.10.5.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+zwarm-3.10.5.dist-info/entry_points.txt,sha256=u0OXq4q8d3yJ3EkUXwZfkS-Y8Lcy0F8cWrcQfoRxM6Q,46
+zwarm-3.10.5.dist-info/RECORD,,