zwarm 3.10.1__tar.gz → 3.10.2__tar.gz

{zwarm-3.10.1 → zwarm-3.10.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: zwarm
-Version: 3.10.1
+Version: 3.10.2
 Summary: Multi-Agent CLI Orchestration Research Platform
 Requires-Python: <3.14,>=3.13
 Requires-Dist: prompt-toolkit>=3.0.52

{zwarm-3.10.1 → zwarm-3.10.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "zwarm"
-version = "3.10.1"
+version = "3.10.2"
 description = "Multi-Agent CLI Orchestration Research Platform"
 readme = "README.md"
 requires-python = ">=3.13,<3.14"

{zwarm-3.10.1 → zwarm-3.10.2}/src/zwarm/prompts/pilot.py

@@ -44,10 +44,13 @@ You command executor agents - capable coding agents that do specific tasks. Thin
 
 # Your Tools
 
-**delegate(task, adapter="codex", model=None, working_dir=None)** - Dispatch a crew member. Returns immediately with session_id.
-  - `adapter`: "codex" (fast, great for code) or "claude" (powerful reasoning)
-  - `model`: Override model (default: gpt-5.1-codex-mini for codex, sonnet for claude)
-  - Use codex for most tasks - it's fast. Use claude for complex reasoning.
+**delegate(task, model=None, working_dir=None)** - Dispatch a crew member. Returns immediately with session_id.
+  - `model`: Just use the model name - adapter is auto-detected:
+    - `"5.2"` or `"gpt-5.2-codex"` - GPT-5.2 Codex (default, fast, great for code)
+    - `"5.2-think"` - GPT-5.2 with extended reasoning
+    - `"opus"` - Claude Opus (most capable, complex reasoning)
+    - `"sonnet"` - Claude Sonnet (balanced)
+  - Use codex models for most tasks - they're fast. Use opus for complex reasoning.
 
 **converse(session_id, message)** - Send follow-up to a crew member. Returns immediately.
 
@@ -55,11 +58,16 @@ You command executor agents - capable coding agents that do specific tasks. Thin
 
 **check_session(session_id)** - Get FULL result. Complete response, tokens, runtime.
 
-**get_trajectory(session_id, full=False)** - See what steps the agent took (for debugging).
+**get_trajectory(session_id, full=False)** - See what steps the agent took.
+  - `full=True`: Show complete untruncated content for all steps (debugging)
+  - `full=False`: Concise summaries (default)
 
-**list_sessions()** - See all crew. `needs_attention=True` means ready for review.
+**list_sessions(status=None)** - See all crew. `needs_attention=True` means ready for review.
+  - `status`: Filter by "running", "completed", "failed", or None for all
 
-**end_session(session_id)** - Dismiss a crew member.
+**end_session(session_id, reason=None, delete=False)** - Dismiss a crew member.
+  - `reason`: Optional note about why
+  - `delete=True`: Permanently remove from list (otherwise just kills if running)
 
 **sleep(seconds)** - Wait before checking. Give crew time to work (15-60s typical).
 
@@ -68,7 +76,7 @@ You command executor agents - capable coding agents that do specific tasks. Thin
 # Workflow
 
 ```
-1. delegate(task) → session_id
+1. delegate(task, model="5.2") → session_id   # or model="opus" for complex tasks
 2. sleep(30)
 3. peek_session(id) → done?
 4. If running, goto 2