feat: add pitrader_script step type + image vision for pipeline triggers

Add pitrader_script executor for running PITrader scripts (pi-scan,
playbook, execute_trades) as pipeline steps with vault credential
injection and JSON output capture.

Extend claude_prompt step with vision support (image_b64 in trigger
context). Add image pipeline trigger to on_image_message handler.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

pipelines/steps/claude_prompt.py

@@ -7,12 +7,17 @@ logger = logging.getLogger(__name__)
 
 async def execute_claude_prompt(
     config: dict,
+    context: dict | None = None,
     llm=None,
     default_model: str = "claude-haiku",
     escalation_model: str = "claude-sonnet",
     **_kwargs,
 ) -> str:
-    """Send a prompt to Claude and return the response."""
+    """Send a prompt to Claude and return the response.
+
+    Supports vision: if config contains 'image_b64' or trigger context has
+    'image_b64', the image is included as a vision content block.
+    """
     if not llm:
         raise RuntimeError("LLM client not configured")
 
@@ -23,9 +28,34 @@ async def execute_claude_prompt(
     model_name = config.get("model", "default")
     model = escalation_model if model_name == "escalation" else default_model
 
+    # Check for image data (from config or trigger context)
+    image_b64 = config.get("image_b64", "")
+    image_mime = config.get("image_mime", "image/png")
+    if not image_b64 and context:
+        trigger = context.get("trigger", {})
+        image_b64 = trigger.get("image_b64", "")
+        image_mime = trigger.get("mime_type", "image/png")
+
+    # Build message content
+    if image_b64:
+        content = [
+            {
+                "type": "image_url",
+                "image_url": {
+                    "url": f"data:{image_mime};base64,{image_b64}",
+                },
+            },
+            {
+                "type": "text",
+                "text": prompt,
+            },
+        ]
+    else:
+        content = prompt
+
     response = await llm.chat.completions.create(
         model=model,
-        messages=[{"role": "user", "content": prompt}],
+        messages=[{"role": "user", "content": content}],
         max_tokens=4096,
     )