Context Curator Administration

Server administration for Context Curator - the intelligent code indexing and context enrichment system.

Quick Status Check

Check all services:

curator-admin status

Expected output:

{
  "curator_api": "healthy",
  "chromadb": "connected",
  "ollama": "ready",
  "sync_agents": 0,
  "last_indexing": "2026-03-08T14:32:15Z",
  "indexed_projects": 3,
  "total_chunks": 1247
}

Administration Commands

1. Check System Status

Full system health check:

#!/bin/bash

echo "=== Context Curator System Status ==="
echo ""

# Check Curator API
echo "1. Curator API Health:"
if curl -s http://127.0.0.1:5555/health > /dev/null 2>&1; then
  echo "   Status: OK"
else
  echo "   Status: FAILED - API not responding"
fi

# Check ChromaDB
echo ""
echo "2. ChromaDB Status:"
if curl -s http://127.0.0.1:8000/api/v1/heartbeat > /dev/null 2>&1; then
  echo "   Status: OK"

  # Get collection stats
  STATS=$(curl -s http://127.0.0.1:8000/api/v1/collections)
  echo "   Collections: $(echo $STATS | jq 'length')"
else
  echo "   Status: FAILED - ChromaDB not responding"
fi

# Check Ollama
echo ""
echo "3. Ollama Models:"
MODELS=$(curl -s http://127.0.0.1:11434/api/tags | jq -r '.models[].name')
if [ -n "$MODELS" ]; then
  echo "   Models loaded:"
  echo "$MODELS" | sed 's/^/     - /'
else
  echo "   Status: FAILED - No models loaded"
fi

# List indexed projects
echo ""
echo "4. Indexed Projects:"
curl -s http://127.0.0.1:5555/projects | jq '.projects[]' 2>/dev/null || echo "   (Unable to fetch)"

# Check sync agents
echo ""
echo "5. Sync Agents:"
AGENTS=$(curl -s http://127.0.0.1:5555/agents 2>/dev/null | jq '.agents | length')
echo "   Connected: $AGENTS"

2. Indexing and Re-indexing

Check current indexing state:

curl -s http://127.0.0.1:5555/projects | jq '.projects[] | {name, status, last_indexed, chunk_count}'

Reindex specific project:

#!/bin/bash

PROJECT="${1:-}"

if [ -z "$PROJECT" ]; then
  echo "Usage: curator reindex <project-path>"
  exit 1
fi

echo "Reindexing: $PROJECT"

# Start reindexing
curl -X POST http://127.0.0.1:5555/projects/reindex \
  -H "Content-Type: application/json" \
  -d "{\"path\": \"$PROJECT\"}" \
  | jq .

# Monitor progress
while true; do
  STATUS=$(curl -s http://127.0.0.1:5555/projects \
    | jq ".projects[] | select(.name == \"$(basename $PROJECT)\") | .status")

  if [[ "$STATUS" == *"completed"* ]]; then
    echo "Reindexing complete!"
    break
  fi

  echo "Status: $STATUS"
  sleep 2
done

Full system reindex:

#!/bin/bash

echo "Starting full system reindex..."

# Reindex all projects
PROJECTS=$(curl -s http://127.0.0.1:5555/projects | jq -r '.projects[].name')

for PROJECT in $PROJECTS; do
  echo "Reindexing: $PROJECT"

  curl -X POST http://127.0.0.1:5555/projects/reindex \
    -H "Content-Type: application/json" \
    -d "{\"name\": \"$PROJECT\"}" > /dev/null

  # Wait for completion
  while [ "$(curl -s http://127.0.0.1:5555/projects | jq ".projects[] | select(.name == \"$PROJECT\") | .status" | grep -c "completed")" -eq 0 ]; do
    sleep 2
  done

  echo "  ✓ Complete"
done

echo "All projects reindexed!"

3. CLAUDE.local.md Generation

Generate CLAUDE.local.md for current project:

#!/bin/bash

PROJECT_PATH="${1:-.}"
OUTPUT="${PROJECT_PATH}/.claude/CLAUDE.local.md"

echo "Generating CLAUDE.local.md..."

mkdir -p "$(dirname "$OUTPUT")"

# Get project metadata from Curator
METADATA=$(curl -s http://127.0.0.1:5555/projects \
  | jq ".projects[] | select(.path == \"$PROJECT_PATH\")")

# Generate header
cat > "$OUTPUT" << 'EOF'
# CLAUDE.local.md

## Project Context

This file is auto-generated by Context Curator.
Last updated: $(date -u +%Y-%m-%dT%H:%M:%SZ)

EOF

# Add metadata
echo "$METADATA" | jq -r '.description' >> "$OUTPUT"

# Add key files
echo "" >> "$OUTPUT"
echo "## Key Files" >> "$OUTPUT"

curl -s "http://127.0.0.1:5555/projects/files" \
  -d "{\"path\": \"$PROJECT_PATH\"}" \
  | jq -r '.files[] | "- \(.path): \(.description)"' >> "$OUTPUT" 2>/dev/null

# Add code patterns
echo "" >> "$OUTPUT"
echo "## Code Patterns" >> "$OUTPUT"

curl -s "http://127.0.0.1:5555/projects/patterns" \
  -d "{\"path\": \"$PROJECT_PATH\"}" \
  | jq -r '.patterns[] | "- \(.name): \(.description)"' >> "$OUTPUT" 2>/dev/null

echo "Generated: $OUTPUT"

Generate full .claude/ OS:

#!/bin/bash

PROJECT_PATH="${1:-.}"
CLAUDE_DIR="${PROJECT_PATH}/.claude"

echo "Regenerating full .claude/ configuration..."

mkdir -p "$CLAUDE_DIR"

# 1. Generate CLAUDE.local.md
echo "1. Generating CLAUDE.local.md..."
curator generate "$PROJECT_PATH" > "$CLAUDE_DIR/CLAUDE.local.md"

# 2. Generate settings.json with context enrichment
echo "2. Generating settings.json..."
cat > "$CLAUDE_DIR/settings.json" << 'EOF'
{
  "version": "1.0.0",
  "context": {
    "maxTokens": 100000,
    "enrichmentLevels": ["local", "project", "semantic"],
    "curator": {
      "enabled": true,
      "apiUrl": "http://127.0.0.1:5555"
    }
  },
  "tools": {
    "allowed": ["Bash", "Read", "Grep", "Glob", "WebSearch"],
    "guards": {
      "bash": "hooks/scripts/bash-guard.sh",
      "write": "hooks/scripts/write-guard.sh"
    }
  }
}
EOF

# 3. Generate memory section
echo "3. Generating memory structure..."
mkdir -p "$CLAUDE_DIR/memory"

cat > "$CLAUDE_DIR/memory/project.md" << 'EOF'
# Project Memory

## Architecture
[Generated from code analysis]

## Key Decisions
[Generated from commit history]

## Performance Baselines
[Generated from profiling data]

## Known Issues
[Generated from issue tracker]
EOF

# 4. Generate agents configuration
echo "4. Generating agents configuration..."
mkdir -p "$CLAUDE_DIR/agents"

cat > "$CLAUDE_DIR/agents/curator.yaml" << 'EOF'
name: "Code Context Agent"
role: "Provide enriched code context"
tools:
  - Read
  - Grep
  - Glob
integrations:
  - curator
  - ollama
EOF

echo "Full .claude/ OS regenerated at: $CLAUDE_DIR"

4. ChromaDB Monitoring

Check ChromaDB health and collections:

#!/bin/bash

echo "=== ChromaDB Monitoring ==="
echo ""

# Heartbeat
echo "1. Heartbeat:"
curl -s http://127.0.0.1:8000/api/v1/heartbeat | jq .

# List collections
echo ""
echo "2. Collections:"
COLLECTIONS=$(curl -s http://127.0.0.1:8000/api/v1/collections)
echo "$COLLECTIONS" | jq '.[] | {name, metadata, count}'

# Collection details
echo ""
echo "3. Collection Stats:"
for COLLECTION in $(echo "$COLLECTIONS" | jq -r '.[].name'); do
  COUNT=$(curl -s "http://127.0.0.1:8000/api/v1/collections/$COLLECTION/count")
  echo "  $COLLECTION: $COUNT documents"
done

# Embedding stats
echo ""
echo "4. Embedding Model:"
EMBEDDING=$(curl -s http://127.0.0.1:8000/api/v1/embedding)
echo "$EMBEDDING" | jq '{model, dimension}'

Optimize ChromaDB:

#!/bin/bash

echo "Optimizing ChromaDB..."

# Compact collections (removes tombstones)
COLLECTIONS=$(curl -s http://127.0.0.1:8000/api/v1/collections | jq -r '.[].name')

for COLLECTION in $COLLECTIONS; do
  echo "Compacting: $COLLECTION"
  curl -X POST "http://127.0.0.1:8000/api/v1/collections/$COLLECTION/compact" \
    -H "Content-Type: application/json"
done

echo "Optimization complete"

Reset ChromaDB (careful!):

#!/bin/bash

echo "WARNING: This will clear all indexed data!"
read -p "Continue? (yes/no) " -n 3 -r
echo
if [[ $REPLY =~ ^[Yy][Ee][Ss]$ ]]; then
  curl -X POST http://127.0.0.1:8000/api/v1/reset \
    -H "Content-Type: application/json"
  echo "ChromaDB reset complete"
else
  echo "Cancelled"
fi

5. Ollama Model Management

List available models:

curl -s http://127.0.0.1:11434/api/tags | jq '.models[] | {name, size}'

Pull model:

ollama pull qwen3-coder
ollama pull nomic-embed-text

Check model status:

ollama list

Sync Agent Management

Check connected sync agents:

curl -s http://127.0.0.1:5555/agents | jq '.agents[] | {id, name, status, last_sync}'

Register new sync agent:

curl -X POST http://127.0.0.1:5555/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "name": "m3-pro-agent",
    "api_key": "key-here",
    "url": "http://REMOTE_HOST:5556"
  }' | jq .

Monitor sync agent:

#!/bin/bash

AGENT_ID="${1:-}"

if [ -z "$AGENT_ID" ]; then
  echo "Usage: curator monitor-agent <agent-id>"
  exit 1
fi

echo "Monitoring agent: $AGENT_ID"

while true; do
  STATUS=$(curl -s http://127.0.0.1:5555/agents/$AGENT_ID)

  echo "$(date): $(echo $STATUS | jq -r '.status')"

  sleep 5
done

Backup and Recovery

Backup ChromaDB:

#!/bin/bash

BACKUP_DIR="/backups/chromadb-$(date +%Y%m%d-%H%M%S)"
mkdir -p "$BACKUP_DIR"

# Stop Curator API to ensure consistency
systemctl stop curator-api

# Copy ChromaDB data
cp -r /var/lib/chromadb/data "$BACKUP_DIR/"

# Restart
systemctl start curator-api

echo "Backup complete: $BACKUP_DIR"

Restore from backup:

#!/bin/bash

BACKUP_DIR="${1:-}"

if [ -z "$BACKUP_DIR" ]; then
  echo "Usage: curator restore <backup-dir>"
  exit 1
fi

echo "Restoring from: $BACKUP_DIR"

# Stop services
systemctl stop curator-api

# Restore data
rm -rf /var/lib/chromadb/data
cp -r "$BACKUP_DIR/data" /var/lib/chromadb/

# Restart
systemctl start curator-api

echo "Restore complete"

Troubleshooting

Curator API not responding:

# Check service status
systemctl status curator-api

# Restart service
systemctl restart curator-api

# Check logs
journalctl -u curator-api -n 50 -f

ChromaDB connection failed:

# Check if ChromaDB is running
ps aux | grep chromadb

# Restart ChromaDB
docker restart chromadb

# Verify connection
curl http://127.0.0.1:8000/api/v1/heartbeat

Ollama models not loading:

# Check Ollama status
curl -s http://127.0.0.1:11434/api/tags

# Restart Ollama
pkill ollama
ollama serve

# Pull required models
ollama pull qwen3-coder
ollama pull nomic-embed-text

Automation Scripts

Daily system check:

#!/bin/bash
# curator-daily-check.sh

TIMESTAMP=$(date -u +%Y-%m-%dT%H:%M:%SZ)

echo "=== Daily Curator Check: $TIMESTAMP ===" >> curator-daily.log

# Health check
curator status >> curator-daily.log

# Reindex if needed
if [ $(date +%H) -eq 2 ]; then
  echo "Running nightly reindex..." >> curator-daily.log
  curator reindex-all >> curator-daily.log
fi

# Optimize
curator optimize >> curator-daily.log

echo "Check complete" >> curator-daily.log

Schedule with cron:

# Run daily at 2 AM
0 2 * * * /usr/local/bin/curator-daily-check.sh

Summary

Context Curator administration includes:

• Status monitoring and health checks
• Project indexing and reindexing
• CLAUDE.md file generation
• ChromaDB management and optimization
• Ollama model management
• Sync agent registration and monitoring
• Backup and recovery procedures
• Automated maintenance tasks