GitHub Sync - Synchronize GitHub Content

Synchronize GitHub issues, pull requests, commits, and CI results from the configured repository into the AI Memory discussions collection.

Usage

# Incremental sync (default) - only fetch updated items
/aim-github-sync

# Full sync - fetch all items from scratch
/aim-github-sync --full

# Check sync status (last sync time, items synced, errors)
/aim-github-sync --status

Options

• --incremental - Sync only updated items since last sync (default)
• --full - Full sync: fetch all items from scratch
• --status - Display sync status (last sync time, items synced per type)

Sync Modes

Incremental Sync (Default)

• Fetches items updated since last sync timestamp per document type
• Faster and more efficient for regular updates
• Priority order: PRs -> Issues -> Commits -> CI Results
• Updates existing documents via dedup/versioning protocol (SPEC-005)

Full Sync

• Fetches all items from scratch
• Use for initial setup or after schema changes
• Warning: Can be slow for large repositories

Configuration

Sync requires GitHub credentials in .env:

GITHUB_TOKEN=ghp_your_token_here
GITHUB_REPO=owner/repo-name
GITHUB_BRANCH=main
GITHUB_SYNC_ENABLED=true

Implementation Reference

This skill invokes GitHubSyncEngine.sync() from src/memory/connectors/github/sync.py:

# The skill invokes the sync engine via a thin script
cd "$AI_MEMORY_INSTALL_DIR" || { echo "Error: AI_MEMORY_INSTALL_DIR is not set or directory does not exist"; exit 1; }
MODE="${MODE:-incremental}"
for arg in "$@"; do
  case "$arg" in
    --full) MODE="full" ;;
    --incremental) MODE="incremental" ;;
  esac
done
"$AI_MEMORY_INSTALL_DIR/.venv/bin/python" -c "
import asyncio
from memory.connectors.github.sync import GitHubSyncEngine
engine = GitHubSyncEngine()
result = asyncio.run(engine.sync(mode='${MODE}'))
print(f'Synced {result.total_synced} items ({result.items_skipped} skipped, {result.errors} errors) in {result.duration_seconds:.1f}s')
d = result.to_dict()
for k, v in d.items():
    if v and k != 'total_synced':
        print(f'  {k}: {v}')
"

Status Mode

Reads .audit/state/github_sync_state.json and displays last sync timestamps per type:

cd "$AI_MEMORY_INSTALL_DIR" || { echo "Error: AI_MEMORY_INSTALL_DIR is not set or directory does not exist"; exit 1; }
python3 -c "
import json
from pathlib import Path
state_file = Path('.audit/state/github_sync_state.json')
if not state_file.exists():
    print('No sync state found. Run /aim-github-sync first.')
else:
    state = json.loads(state_file.read_text())
    print('## GitHub Sync Status')
    print('')
    for type_key, info in state.items():
        last = info.get('last_synced', 'never')
        count = info.get('last_count', 0)
        print(f'  {type_key}: last synced {last} ({count} items)')
"

Guard

If GITHUB_SYNC_ENABLED is not true, the skill will display:

Error: GitHub sync is not enabled. Set GITHUB_SYNC_ENABLED=true and configure GITHUB_TOKEN and GITHUB_REPO.

Technical Details

• Rate Limiting: Uses GitHubClient with configurable request delays
• Pagination: Automatic pagination for all GitHub API endpoints
• Deduplication: SHA256 content hashing prevents duplicate storage (SPEC-005)
• Versioning: Changed content creates new version, marks old as superseded
• Collection: discussions (shared with conversation data, filtered by source="github")
• Tenant Isolation: group_id = owner/repo
• Sync Priority: PRs (+ reviews + diffs) -> Issues (+ comments) -> Commits -> CI Results

Notes

• Requires GitHub personal access token (classic or fine-grained)
• Sync logs written to ~/.ai-memory/logs/activity.log
• First full sync can take several minutes for large repositories
• Incremental sync timestamps stored per type in github_sync_state.json
• Reviews and diffs are synced as part of PR sync
• Issue comments are synced as part of issue sync