story-kit: create 158_bug_pty_debug_log_panics_on_multi_byte_utf_8_characters

.story_kit/work/1_upcoming/158_bug_pty_debug_log_panics_on_multi_byte_utf_8_characters.md

@@ -0,0 +1,28 @@
+---
+name: "PTY debug log panics on multi-byte UTF-8 characters"
+---
+
+# Bug 158: PTY debug log panics on multi-byte UTF-8 characters
+
+## Description
+
+The PTY debug logging in `claude_code.rs` uses byte-level string slicing (`&trimmed[..trimmed.len().min(120)]`) which panics when byte 120 falls inside a multi-byte UTF-8 character like an em dash (`—`, 3 bytes: E2 80 94).
+
+## How to Reproduce
+
+1. Start an agent on a story
+2. Have the agent process a log file or content that causes Claude to output an em dash (`—`) near the 120-byte boundary of a JSON stream event line
+3. The PTY task panics with "byte index 120 is not a char boundary"
+
+## Actual Result
+
+WebSocket error: PTY task panicked with "byte index 120 is not a char boundary; it is inside '—' (bytes 118..121)"
+
+## Expected Result
+
+The debug log should safely truncate the string at a valid UTF-8 char boundary without panicking.
+
+## Acceptance Criteria
+
+- [ ] Replace `&trimmed[..trimmed.len().min(120)]` with `&trimmed[..trimmed.floor_char_boundary(120)]` in `server/src/llm/providers/claude_code.rs:251`
+- [ ] Agent sessions no longer panic when Claude outputs multi-byte UTF-8 characters near the truncation boundary