Use Claude Code, Codex, VSCode, Pi, and OpenCode (and 6 other harnesses) for free (1.3B+ free tokens) from your terminal, app, IDE, or phone, and now from the browser with native browser sessions (multi-harness + multi-model) like OpenClaw (voice supported + ToS friendly)
Python
switched to start truncation from end truncation
Alishahryar107b0c9233e9d
diff --git a/messaging/handler.py b/messaging/handler.pyindex 7ac54d83..3bd7908e 100644--- a/messaging/handler.py+++ b/messaging/handler.py@@ -384,7 +384,7 @@ class ClaudeMessageHandler:thinking_text = "".join(components["thinking"])# Truncate thinking if too long, it's usually less critical than final contentif len(thinking_text) > 1000:- thinking_text = thinking_text[:995] + "..."+ thinking_text = "..." + thinking_text[-995:]# Ensure it doesn't break a code block if we eventually support them inside thinkinglines.append(f"💭 **Thinking:**\n```\n{thinking_text}\n```")@@ -435,13 +435,16 @@ class ClaudeMessageHandler:else "⏳ **Claude is working...**")- # If too long, truncate the main content but keep the status and close code blocks+ # If too long, truncate the start of the content (keep the end)available_limit = LIMIT - len(status_text) - 20 # 20 for truncation marker- truncated_main = main_text[:available_limit].rstrip() + "\n... (truncated)"+ raw_truncated = main_text[-available_limit:].lstrip()- # Close any open code blocks- if truncated_main.count("```") % 2 != 0:- truncated_main += "\n```"+ # Check for unbalanced code blocks+ prefix = "... (truncated)\n"+ if raw_truncated.count("```") % 2 != 0:+ prefix += "```\n"++ truncated_main = prefix + raw_truncatedreturn truncated_main + status_text