Manage SpecMem service modes, file watcher, and check connectivity.

COMMAND PARSING

Parse what comes after /specmem-service:

ARGUMENTS: $ARGUMENTS

DECISION LOGIC:

If ARGUMENTS is empty, blank, or "help" -> Execute HELP section
If ARGUMENTS is "status" -> Execute STATUS section
If ARGUMENTS is "start" -> Execute START section
If ARGUMENTS is "stop" -> Execute STOP section
If ARGUMENTS is "restart" -> Execute RESTART section
If ARGUMENTS is "mcp" -> Execute MCP section
If ARGUMENTS is "http" -> Execute HTTP section
If ARGUMENTS is "db" -> Execute DB section
If ARGUMENTS is "watch" or "watch start" -> Execute WATCH_START section
If ARGUMENTS is "watch stop" -> Execute WATCH_STOP section
If ARGUMENTS is "watch status" -> Execute WATCH_STATUS section
If ARGUMENTS is "sync" or "sync check" -> Execute SYNC_CHECK section
If ARGUMENTS is "sync force" or "resync" -> Execute SYNC_FORCE section
Otherwise -> Execute HELP section (unknown command)

HELP

Show this help output and STOP:

SpecMem Service - Mode Management

USAGE:
  /specmem-service              Show this help
  /specmem-service status       Check all service connectivity
  /specmem-service start        Start SpecMem services
  /specmem-service stop         Stop SpecMem services
  /specmem-service restart      Restart SpecMem services
  /specmem-service mcp          Test MCP tool availability
  /specmem-service http         Test HTTP API endpoint
  /specmem-service db           Test database connection

FILE WATCHER:
  /specmem-service watch        Start file watcher (auto-sync code changes)
  /specmem-service watch start  Start file watcher
  /specmem-service watch stop   Stop file watcher
  /specmem-service watch status Check watcher status

SYNC MANAGEMENT:
  /specmem-service sync         Check if files are in sync
  /specmem-service sync check   Check sync status (detailed)
  /specmem-service sync force   Force full resync
  /specmem-service resync       Force full resync (alias)

SERVICE MODES:
  MCP     Standard MCP via stdio (native Claude integration)
  HTTP    REST API at http://localhost:8595
  DB      Direct PostgreSQL connection
  WATCH   File watcher for auto-sync

PORTS:
  8595    Dashboard/HTTP API
  5432    PostgreSQL database
  3456    Embedding service

RELATED:
  /specmem-stats    View system statistics
  /specmem          Main SpecMem commands

STATUS

Check all service connectivity:

Step 1 - Check MCP: Call the MCP tool mcp__specmem__show_me_the_stats with empty parameters {}.

If the tool returns successfully -> Report: "MCP: CONNECTED" If the tool fails or errors -> Report: "MCP: NOT AVAILABLE"

Step 2 - Check HTTP: Execute this bash command:

curl -s -o /dev/null -w "%{http_code}" http://localhost:8595/api/health 2>/dev/null || echo "000"

If output is "200" -> Report: "HTTP: CONNECTED (port 8595)" Otherwise -> Report: "HTTP: NOT AVAILABLE"

Step 3 - Check Database: Execute this bash command:

pg_isready -h localhost -p 5432 2>/dev/null && echo "ready" || echo "not ready"

If output contains "ready" (not "not ready") -> Report: "DB: CONNECTED (port 5432)" Otherwise -> Report: "DB: NOT AVAILABLE"

Step 4 - Check File Watcher: Call the MCP tool mcp__specmem__check_sync_status with parameters:

{
  "detailed": false
}

If the tool returns successfully and watcherStatus.isRunning is true -> Report: "WATCHER: RUNNING (syncing code changes)" If the tool returns successfully and watcherStatus.isRunning is false -> Report: "WATCHER: STOPPED" Otherwise -> Report: "WATCHER: UNKNOWN"

Step 5 - Display Summary: Show all status results in a formatted summary.

START

Start SpecMem services.

Execute this bash command:

cd /specmem && if [ -f start-specmem.sh ]; then echo "Starting SpecMem services..." && bash start-specmem.sh; elif [ -f start-dashboard.sh ]; then echo "Starting dashboard..." && bash start-dashboard.sh; else echo "ERROR: No startup script found at /specmem"; exit 1; fi

Report the result to the user.

STOP

Stop SpecMem services.

Execute this bash command:

cd /specmem && if [ -f stop-specmem.sh ]; then echo "Stopping SpecMem services..." && bash stop-specmem.sh; else echo "Stopping services via pkill..." && pkill -f "specmem" 2>/dev/null || true && pkill -f "node.*specmem" 2>/dev/null || true && echo "Services stopped (or were not running)"; fi

Report the result to the user.

RESTART

Restart SpecMem services.

Execute this bash command:

cd /specmem && echo "Restarting SpecMem services..." && (if [ -f stop-specmem.sh ]; then bash stop-specmem.sh; else pkill -f "specmem" 2>/dev/null || true; fi) && sleep 2 && (if [ -f start-specmem.sh ]; then bash start-specmem.sh; elif [ -f start-dashboard.sh ]; then bash start-dashboard.sh; else echo "ERROR: No startup script found"; exit 1; fi)

Report the result to the user.

MCP

Test MCP tool availability.

Call the MCP tool mcp__specmem__show_me_the_stats with parameters:

{
  "includeTypeDistribution": true,
  "includeImportanceDistribution": true
}

Display the result showing memory counts and distributions. If the call fails, report that MCP is not available.

HTTP

Test HTTP API endpoint.

Execute this bash command:

echo "=== HTTP API CHECK ===" && echo "" && echo "Endpoint: http://localhost:8595" && echo "" && echo "Health Check:" && curl -s http://localhost:8595/api/health 2>/dev/null || echo "  (not responding)" && echo "" && echo "Available Endpoints:" && echo "  GET  /api/health" && echo "  GET  /api/stats" && echo "  GET  /api/memories" && echo "  GET  /api/team-members/active" && echo "  POST /api/memories/search"

Report the output to the user.

DB

Test database connection.

Execute this bash command:

echo "=== DATABASE CHECK ===" && echo "" && echo "Host: localhost:5432" && echo "" && pg_isready -h localhost -p 5432 2>/dev/null || echo "Database not ready" && echo "" && echo "Tables (if connected):" && PGPASSWORD="${SPECMEM_DB_PASSWORD:-${SPECMEM_PASSWORD:-specmem_westayunprofessional}}" psql -h localhost -p 5432 -U "${SPECMEM_DB_USER:-specmem_westayunprofessional}" -d "${SPECMEM_DB_NAME:-specmem_westayunprofessional}" -c "SELECT tablename FROM pg_tables WHERE schemaname='public' LIMIT 10;" 2>/dev/null || echo "  (cannot connect)"

Report the output to the user.

WATCH_START

Start the file watcher to auto-sync code changes to memories.

Call the MCP tool mcp__specmem__start_watching with parameters:

{
  "syncCheckIntervalMinutes": 60
}

Display the result to the user:

If successful: Show projectPath, watchedPaths, filesWatched count
If already running: Indicate it's already running and show stats
If error: Show error message

Notes:

File watcher is PROJECT-SCOPED (only watches current project)
rootPath parameter is DEPRECATED (always uses SPECMEM_PROJECT_PATH)
Default sync check interval is 60 minutes (1 hour)
Keeps MCP memories in sync with filesystem automatically

WATCH_STOP

Stop the file watcher.

Call the MCP tool mcp__specmem__stop_watching with parameters:

{
  "flushPending": true
}

Display the result to the user:

If successful: Show eventsProcessed and pendingFlushed counts
If not running: Indicate watcher wasn't running
If error: Show error message

Notes:

flushPending: true processes all pending changes before stopping
flushPending: false stops immediately without processing queue

WATCH_STATUS

Check file watcher status.

Call the MCP tool mcp__specmem__check_sync_status with parameters:

{
  "detailed": false
}

Display the result to the user:

Watcher status: isRunning, eventsProcessed, queueSize
Sync status: inSync, syncScore, driftPercentage
Summary message
Stats: totalFiles, totalMemories, upToDate, missingFromMcp, missingFromDisk, contentMismatch

If watcher is not running, suggest starting it with /specmem-service watch start.

SYNC_CHECK

Check if MCP memories are in sync with filesystem.

Call the MCP tool mcp__specmem__check_sync_status with parameters:

{
  "detailed": true
}

Display the result to the user:

Sync status: inSync (true/false)
Sync score: percentage (0-100%)
Drift percentage: how much drift detected
Summary message
Statistics: totalFiles, totalMemories, upToDate, missingFromMcp, missingFromDisk, contentMismatch
Detailed lists (because detailed: true):
- Files missing from MCP
- Files deleted from disk
- Files with content mismatch

If drift is detected, suggest running /specmem-service sync force to resync.

Notes:

This is a read-only check (doesn't modify anything)
Shows what's out of sync between filesystem and memories
Use this before force_resync to see what will be changed

SYNC_FORCE

Force a full resync of the entire codebase.

Step 1 - Dry Run Preview: Call the MCP tool mcp__specmem__force_resync with parameters:

{
  "dryRun": true
}

Display the preview to the user:

Files that would be added
Files that would be updated
Files that would be marked deleted

Step 2 - Ask for Confirmation: Ask the user: "Do you want to proceed with the resync? This will update all MCP memories to match the current filesystem. (yes/no)"

Wait for user response.

Step 3 - Execute Resync (if confirmed): If user confirms (yes/y/proceed/go/confirm): Call the MCP tool mcp__specmem__force_resync with parameters:

{
  "dryRun": false
}

Display the result:

Success/failure status
Files added
Files updated
Files marked deleted
Duration (milliseconds)
Any errors encountered

If user declines: Report: "Resync cancelled by user."

Notes:

Always shows preview first (dry run)
Requires explicit user confirmation
Scans entire codebase and updates all memories
Use after: git checkout, git pull, mass file operations
Can take several seconds to minutes on large codebases

SpecMem Service - Service Mode Management

COMMAND PARSING

HELP

STATUS

START

STOP

RESTART

MCP

HTTP

DB

WATCH_START

WATCH_STOP

WATCH_STATUS

SYNC_CHECK

SYNC_FORCE