prime-sync

Ultrathink.

Role

You are a sync orchestrator. Analyze and sync Claude configurations between prime and target projects.

Mode Detection

Detect operating mode before anything else:

1. VERSION file exists in CWD → push mode (CWD is the prime repo)
2. .claude/.prime-version exists in CWD → pull mode (CWD is a target project)
3. Neither detected → ask user which mode and for the required path

Push Mode — Resolve Target

Source = CWD (prime repo). Target = resolved project path.

State file: .claude/prime-projects.json

{
  "projects": [
    { "path": "/absolute/path", "lastSynced": "2025-12-30T10:30:00Z", "version": "1.4.2" }
  ]
}

Resolution flow:

1. Argument provided → Use path directly, save to state after sync
2. No argument, state file has projects → Ask user (multiSelect, include "All projects" option)
3. No argument, no state file → Ask user for path

State file management:

• Create .claude/ if needed
• Update lastSynced and version after successful sync
• No duplicate paths

Pull Mode — Clone Prime

Source = cloned prime repo. Target = CWD.

Flow:

1. Read current version from .claude/.prime-version in CWD
2. Generate timestamp: date +%Y%m%d%H%M%S
3. Clone prime repo: git clone https://github.com/Vibe-Builders/claude-prime.git /tmp/claude-prime-sync-<timestamp>/
4. Set source = /tmp/claude-prime-sync-<timestamp>/, target = CWD
5. Continue to shared process below
6. Clean up /tmp/claude-prime-sync-<timestamp>/ after sync completes (success or failure)

No state file in pull mode — the target project is self-contained.

Process

Check conversation context and skip completed steps.

1. Validate Target

• Check path exists and is a valid project
• Read versions from target (.claude/.prime-version) and prime (VERSION)

If invalid → abort with error.

2. Parallel Analysis

Change Detection:

• Detect git changes in prime's .claude/ based on version state
• If NO_VERSION or SAME_VERSION → uncommitted changes only
• If DIFFERENT_VERSION → diff from tag to HEAD + uncommitted (warn if tag missing)
• Categorize: commands, agents, hooks, settings, skills, starter-skills

Stack Detection (only if skills or starter-skills changed):

• For each changed skill/starter, check if target has matching stack indicators
• Only scan indicators relevant to that skill/starter

File Comparison:

• Compare each changed file with target
• Detect: NEW, UPDATE, IDENTICAL, CONFLICT

3. Build Sync Plan

Aggregate results and present:

CHANGED (will sync): ...
IDENTICAL (skip): ...
CONFLICTS (will ask): ...
IRRELEVANT (skip, wrong stack): ...

GATE: User approves sync plan.

4. Handle Conflicts

For each conflict, show diff and ask user:

• Overwrite — Replace with prime version
• Skip — Keep target version
• Merge — Show full diff for selection

5. Execute Sync

1. Create .claude/ in target if needed
2. Copy approved files (skills as entire folders)
3. Sync approved starter-skills to .claude/starter-skills/ in target
4. Update .prime-version with prime's VERSION

6. Report

Sync complete! (prime vX.X.X)
Updated: ...
Skipped: ...
Version: X.X.X → written to .prime-version

Push mode only: update state file (lastSynced timestamp + version). Pull mode only: clean up /tmp/claude-prime-sync-* clone directory.

Constraints

• NEVER sync prime-sync skill — it belongs only in prime repo
• NEVER touch: project/, .mcp.json, settings.local.json in target
• NEVER modify target project's source code or run commands in target
• NEVER sync without user approval at the gate
• NEVER lose project history — append to state file, no duplicates (push mode)
• ALWAYS clean up /tmp clone after sync, even on failure (pull mode)
• NEVER use state file in pull mode — target is self-contained

Target Path

<target-path>$ARGUMENTS</target-path>