Agent-Skills.md

Agent Skills: Config Sync Push

Use when the user wants to back up, save, or push their current Claude Code config to GitHub.

UncategorizedID: jeremylongshore/claude-code-plugins/sync-push

Repository

jeremylongshore/claude-code-plugins
jeremylongshoreLicense: MIT
2,458348

Install this agent skill to your local

pnpm dlx add-skill https://github.com/jeremylongshore/claude-code-plugins-plus-skills/tree/HEAD/plugins/productivity/claudebase/skills/sync-push

Skill Files

Browse the full folder contents for sync-push.

Download Skill

Loading file tree…

plugins/productivity/claudebase/skills/sync-push/SKILL.md

Skill Metadata

Name
sync-push
Description
Use when the user wants to back up, save, or push their current Claude Code config to GitHub.

Config Sync Push

Push your current local Claude Code configuration to the GitHub backup repo.

What gets pushed

All syncable files from the current project and global config:

  • .mcp.json, .claude/settings.json
  • .claude/agents/, .claude/commands/, .claude/skills/
  • .claude/hooks/ (scripts, config, sounds)
  • .claude/rules/, .claude/agent-memory/
  • .auto-memory/ (memory files)
  • ~/.claude/settings.json (global settings, opt-in via --include-global)

Files that are never pushed: CLAUDE.md (version-controlled with project), settings.local.json, hooks-config.local.json, conversations, sessions.

Instructions

Parse user arguments and run the push script:

bash "${CLAUDE_PLUGIN_ROOT}/scripts/sync-push.sh" [OPTIONS]

Options:

  • --profile NAME — Push to a specific profile (default: active profile)
  • --force — Skip multi-machine safety check and secret warnings
  • --dry-run — Show what would be pushed without actually pushing
  • --include-global — Also push ~/.claude/settings.json (global settings are opt-in)

Multi-machine warning

If the script warns about another machine having pushed more recently, advise the user to pull first (/sync-pull) or use --force if they're sure.

Secret detection

If the script warns about potential secrets, tell the user which files were flagged and suggest reviewing them before using --force.

User Arguments

$ARGUMENTS