feat: add ephemeral prefill messages and system prompt loading

- Implemented functionality to load ephemeral prefill messages from a JSON file, enhancing few-shot priming capabilities for the agent.
- Introduced a mechanism to load an ephemeral system prompt from environment variables or configuration files, ensuring dynamic prompt adjustments at API-call time.
- Updated the CLI and agent initialization to utilize the new prefill messages and system prompt, improving the overall interaction experience.
- Enhanced configuration options with new environment variables for prefill messages and system prompts, allowing for greater customization without persistence.

tools/file_operations.py

@@ -412,9 +412,17 @@ class ShellFileOperations(FileOperations):
             # Still try to read, but warn
             pass
         
-        # Check if it's an image - return base64
+        # Images are never inlined — redirect to the vision tool
         if self._is_image(path):
-            return self._read_image(path)
+            return ReadResult(
+                is_image=True,
+                is_binary=True,
+                file_size=file_size,
+                hint=(
+                    "Image file detected. Automatically redirected to vision_analyze tool. "
+                    "Use vision_analyze with this file path to inspect the image contents."
+                ),
+            )
         
         # Read a sample to check for binary content
         sample_cmd = f"head -c 1000 {self._escape_shell_arg(path)} 2>/dev/null"
@@ -457,6 +465,10 @@ class ShellFileOperations(FileOperations):
             hint=hint
         )
     
+    # Images larger than this are too expensive to inline as base64 in the
+    # conversation context. Return metadata only and suggest vision_analyze.
+    MAX_IMAGE_BYTES = 512 * 1024  # 512 KB
+
     def _read_image(self, path: str) -> ReadResult:
         """Read an image file, returning base64 content."""
         # Get file size
@@ -467,6 +479,17 @@ class ShellFileOperations(FileOperations):
         except ValueError:
             file_size = 0
         
+        if file_size > self.MAX_IMAGE_BYTES:
+            return ReadResult(
+                is_image=True,
+                is_binary=True,
+                file_size=file_size,
+                hint=(
+                    f"Image is too large to inline ({file_size:,} bytes). "
+                    "Use vision_analyze to inspect the image, or reference it by path."
+                ),
+            )
+        
         # Get base64 content
         b64_cmd = f"base64 -w 0 {self._escape_shell_arg(path)} 2>/dev/null"
         b64_result = self._exec(b64_cmd, timeout=30)

tools/file_tools.py

@@ -199,7 +199,7 @@ def _check_file_reqs():
 
 READ_FILE_SCHEMA = {
     "name": "read_file",
-    "description": "Read a file with line numbers and pagination. Use this instead of cat/head/tail in terminal. Output format: 'LINE_NUM|CONTENT'. Suggests similar filenames if not found. Images (png/jpg/gif/webp) returned as base64. Use offset and limit for large files.",
+    "description": "Read a text file with line numbers and pagination. Use this instead of cat/head/tail in terminal. Output format: 'LINE_NUM|CONTENT'. Suggests similar filenames if not found. Use offset and limit for large files. NOTE: Cannot read images or binary files — use vision_analyze for images.",
     "parameters": {
         "type": "object",
         "properties": {

tools/web_tools.py

@@ -159,7 +159,7 @@ async def process_content_with_llm(
         return processed_content
         
     except Exception as e:
-        logger.error("Error processing content with LLM: %s", e)
+        logger.debug("Error processing content with LLM: %s", e)
         return f"[Failed to process content: {str(e)[:100]}. Content size: {len(content):,} chars]"
 
 
@@ -318,7 +318,7 @@ async def _process_large_content_chunked(
             summaries.append(f"## Section {chunk_idx + 1}\n{summary}")
     
     if not summaries:
-        logger.error("All chunk summarizations failed")
+        logger.debug("All chunk summarizations failed")
         return "[Failed to process large content: all chunk summarizations failed]"
     
     logger.info("Got %d/%d chunk summaries", len(summaries), len(chunks))
@@ -532,7 +532,7 @@ def web_search_tool(query: str, limit: int = 5) -> str:
         
     except Exception as e:
         error_msg = f"Error searching web: {str(e)}"
-        logger.error("%s", error_msg)
+        logger.debug("%s", error_msg)
         
         debug_call_data["error"] = error_msg
         _debug.log_call("web_search_tool", debug_call_data)
@@ -673,7 +673,7 @@ async def web_extract_tool(
                 })
                 
             except Exception as scrape_err:
-                logger.error("Error scraping %s: %s", url, scrape_err)
+                logger.debug("Scrape failed for %s: %s", url, scrape_err)
                 results.append({
                     "url": url,
                     "title": "",
@@ -799,7 +799,7 @@ async def web_extract_tool(
             
     except Exception as e:
         error_msg = f"Error extracting content: {str(e)}"
-        logger.error("%s", error_msg)
+        logger.debug("%s", error_msg)
         
         debug_call_data["error"] = error_msg
         _debug.log_call("web_extract_tool", debug_call_data)
@@ -892,7 +892,7 @@ async def web_crawl_tool(
                 **crawl_params
             )
         except Exception as e:
-            logger.error("Crawl API call failed: %s", e)
+            logger.debug("Crawl API call failed: %s", e)
             raise
 
         pages: List[Dict[str, Any]] = []
@@ -1092,7 +1092,7 @@ async def web_crawl_tool(
         
     except Exception as e:
         error_msg = f"Error crawling website: {str(e)}"
-        logger.error("%s", error_msg)
+        logger.debug("%s", error_msg)
         
         debug_call_data["error"] = error_msg
         _debug.log_call("web_crawl_tool", debug_call_data)
@@ -1227,7 +1227,7 @@ WEB_SEARCH_SCHEMA = {
 
 WEB_EXTRACT_SCHEMA = {
     "name": "web_extract",
-    "description": "Extract content from web page URLs. Pages under 5000 chars return raw content; larger pages are LLM-summarized and capped at ~5000 chars per page. Pages over 2M chars are refused. Use browser tools only when pages require interaction or dynamic content.",
+    "description": "Extract content from web page URLs. Returns page content in markdown format. Pages under 5000 chars return full markdown; larger pages are LLM-summarized and capped at ~5000 chars per page. Pages over 2M chars are refused. If a URL fails or times out, use the browser tool to access it instead.",
     "parameters": {
         "type": "object",
         "properties": {