Source Intelligence

DisclaimerUnofficial, and not affiliated with Anthropic. Nearly all of this is read straight out of what ships: npm bundles, captured prompts, published docs. Anthropic's own notes go in verbatim, marked as theirs. The rest is my reading, and every entry carries the strings behind it. If one looks wrong, vote it down and say why.

All of v1.0.77 Home All releases olderv1.0.76 v1.0.78newer
Claude Code v1.0.77

Advanced Shell Command Escaping

What: Sophisticated detection and escaping of dangerous shell constructs like heredocs and multiline strings How to use: This protection is automatic when Claude executes shell commands. It prevents injection attacks in complex scenarios:

# Heredocs are now safely handled
cat << EOF
$(potentially_dangerous_command)
EOF

# Multiline strings are properly escaped
echo 'line1
line2'

Details:

  • Detects heredoc patterns that could bypass normal escaping
  • Identifies multiline strings that might hide malicious commands
  • Applies context-aware escaping strategies
  • Automatically adds stdin redirection (< /dev/null) to prevent interactive command exploitation
  • Protects against arithmetic expansion and nested command substitution attacks

See this entry in the whole of v1.0.77 →