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 releases Home olderv1.0.111 v1.0.113newer

Claude Code v1.0.112

9 entries read diff v1.0.111 → v1.0.112 Markdown

Version 1.0.112 introduces skill.md file support for organizing commands hierarchically, adds enhanced network sandboxing with allow-lists, and removes unused internal network sandboxing code. Several improvements enhance command parsing, error tracking, and user configuration options.

Find
Pick an entry · j / k steps through
9 entries

Changesopen

skill.md File Support#

What: Organize multiple commands in a directory under a single skill.md file How to use:

# Create a directory with related commands
mkdir my-commands/
echo "# My Skill" > my-commands/skill.md
echo "command content" > my-commands/command1.md

# Claude Code will now treat all .md files in this directory as part of the skill

Details:

  • When multiple command files exist in a directory with a skill.md file, Claude Code prioritizes the skill.md
  • Case-insensitive matching (skill.md, Skill.md, SKILL.MD all work)
  • Helps organize complex command structures hierarchically
  • Evidence: uM0() at line 424817 and pZ5() at line 424820 in v1.0.112

Enhanced Network Sandboxing with Allow-Lists#

What: Fine-grained network access control for sandboxed bash commands How to use:

# Network sandboxing now supports allow-lists for specific hosts
# Configured through internal settings: sandbox.network.allow

Details:

  • NetworkManager filters connections based on configured allow-list
  • Blocks all network access except explicitly allowed hosts
  • Logs allowed and blocked connections for transparency
  • Works with HTTP proxy for controlled network access
  • Evidence: NetworkManager implementation at line 373982-374129 in v1.0.112

Automatic Feature Flag Updates#

What: Statsig feature flags now refresh automatically every 6 hours Details:

  • Keeps feature flags up-to-date without restarting the CLI
  • Runs in background, transparent to users
  • Automatically disabled when using Bedrock/Vertex or when telemetry is disabled
  • Evidence: xqB() at line 428347 in v1.0.112

Configurable Spinner Tips#

What changed: Added ability to disable spinner tips through configuration How it works now:

# Configure in settings to disable tips during long operations
# Set spinnerTipsEnabled: false in configuration

Evidence: NrB() at line 437820

Enhanced Command Parsing and Validation#

What changed: Better error handling and validation before command execution Benefits:

  • Clearer error message: "Command cannot be parsed" for malformed commands
  • Improved detection of input redirection patterns (e.g., < file.txt)
  • More robust tokenization of complex commands with special characters
  • Evidence: lAB() at line 384563, qz0() at line 373516, ut2() at line 373524

Better Error Tracking#

What changed: Error reports now include unique error IDs Benefits:

  • Easier to track and debug specific errors when reporting issues
  • Better correlation between user reports and backend logs
  • Evidence: Rh1() at line 428069 and U1() at line 428104

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Improved Native Installation Detection#

What changed: Prevents auto-migration prompts for users already on native installations Benefits:

  • No more unnecessary migration notifications
  • Cleaner user experience for native install users
  • Evidence: Lm1() at line 435204

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Streamlined Terminal Output#

What changed: More consistent terminal output rendering Benefits:

  • Fewer visual glitches in terminal display
  • Better structured output handling
  • Evidence: IC9 variable at line 361107

Related

Other releases about the same thing. Found by shared names or similar wording; neither means one caused the other.

Internal Network Sandboxing (macOS sandbox-exec)#

What was removed: Unused internal network sandboxing implementation Details:

  • Removed ST6() (sandbox-exec profile generator) at line 373486 in v1.0.111
  • Removed ct2() (macOS command wrapper) at line 373499 in v1.0.111
  • Removed at2() (platform-agnostic sandbox orchestrator) at line 373547 in v1.0.111
  • This was internal-only functionality not exposed to users
  • Replaced by the enhanced NetworkManager implementation
Verbatim
Official · Anthropic

Anthropic’s official release notes

Published verbatim by Anthropic for v1.0.112. Text is unmodified from the upstream changelog. Everything else on this page came out of the bundle instead, which is why the two lists don't match.

  • Transcript mode (Ctrl+R): Added the model used to generate each assistant message
  • Addressed issue where some Claude Max users were incorrectly recognized as Claude Pro users
  • Hooks: Added systemMessage support for SessionEnd hooks
  • Added spinnerTipsEnabled setting to disable spinner tips
  • IDE: Various improvements and bug fixes
System prompt

The system prompt was not captured for this release, so this page cannot say whether it moved.