claude-patch

Community-contributed binary patches for Claude Code CLI — unlock hard-coded limitations without waiting for upstream changes.

Disclaimer: This is unofficial and not supported by Anthropic. It modifies the Claude Code binary. Use at your own risk. Keep backups and know how to reinstall Claude Code.

Install

Option A: Agent Skills CLI (works with Claude Code, Cursor, Copilot, etc.)

npx skills add huybuidac/claude-code-patchkit -g

The -g flag installs globally (user-level) so the skill is available in all projects. Without it, the skill is only available in the current project.

Option B: Claude Code plugin system

/plugin marketplace add huybuidac/claude-code-patchkit
/plugin install claude-patch

Prerequisites

• macOS (arm64 tested, x86_64 expected to work)
• Python 3
• codesign (ships with macOS)
• Claude Code installed as native binary
• Write permission to the Claude binary

Usage

Inside a Claude Code session:

/claude-patch                          # Interactive mode
/claude-patch apply subagent-model     # Apply specific patch
/claude-patch revert subagent-model    # Revert to backup
/claude-patch status                   # Show all patch states

Available Patches

Patch	Description	Risk
subagent-model	Unlock model param on Agent tool — use any model id per-call	Low

How It Works

Claude Code is distributed as a bun-compiled binary with embedded JS bundles. Some behaviors are locked behind hard-coded Zod schemas. This is an Agent Skill — Claude reads the patch definitions and executes the procedure interactively with your confirmation at each step. It applies safe, length-preserving binary patches with:

1. Fingerprint detection — locate exact byte patterns
2. Context guard — verify surrounding stable strings (not minified variables)
3. State detection — determine unpatched / patched / abnormal before acting
4. Backup — always saves <binary>.bak.<timestamp>
5. Length-preserving swap — no offset shifts or blob corruption
6. Re-sign — ad-hoc codesign for macOS
7. Self-verify — confirm marker count post-patch

Safety

• Never patches without explicit user confirmation
• Aborts if fingerprint doesn't match (schema changed in new version)
• Always creates backup before modifying
• Re-applies required per Claude Code update (new binary = clean slate)

Contributing

1. Copy skills/claude-patch/patches/TEMPLATE.md
2. Fill in all sections (fingerprint, context guard, replacement, detection, script)
3. Test on at least 2 Claude Code versions
4. Submit a PR

Requirements

• Length-preserving replacement (old bytes == new bytes in length)
• Unique marker string embedded in replacement for state detection
• Context guard using stable strings (.describe() text, not minified variable names)
• Tested version list

Platform Support

• macOS arm64 (Apple Silicon) — tested
• macOS x86_64 — should work (untested)
• Linux — binary layout may differ, contributions welcome

Recovery

• Backups are saved as <binary>.bak.<timestamp> next to the original
• To manually restore: cp <binary>.bak.<timestamp> <binary> && codesign --force --sign - <binary>
• To reinstall Claude Code: delete ~/.local/share/claude/ and re-run the installer

License

MIT