code-helper

/icdev-knowledge — Knowledge Base Management

Usage

/icdev-knowledge <action> [options]

Actions:
  search <query>           Search for patterns and solutions
  add <name>               Add a new pattern to the knowledge base
  recommend <project-id>   Get improvement recommendations
  analyze <error-message>  Analyze a failure for root cause
  stats                    Show knowledge base statistics

What This Does

Manages the self-learning knowledge base that powers self-healing:

1. Search for known patterns, solutions, and best practices
2. Add new patterns discovered during development/operations
3. Get project-specific improvement recommendations
4. Analyze failures to determine root cause
5. View knowledge base statistics and health

Steps

Action: search

Use the search_knowledge MCP tool from icdev-knowledge:

• query: search terms from $ARGUMENTS
• pattern_type: optional filter (error, performance, security, compliance, deployment, configuration)
• limit: default 10

Display matching patterns with:

• Name, type, description
• Detection rule
• Solution
• Confidence score
• Use count

Action: add

Use the add_pattern MCP tool from icdev-knowledge:

• name: from arguments
• Interactive prompts for:
  • pattern_type (error/performance/security/compliance/deployment/configuration)
  • description
  • detection_rule (regex, log pattern, or metric threshold)
  • solution (remediation steps)
  • auto_healable (true/false)
  • confidence (0.0-1.0)

Action: recommend

Use the get_recommendations MCP tool from icdev-knowledge:

• project_id: from arguments
• Returns recommendations based on:
  • Recent failure history
  • Common error patterns
  • Knowledge base pattern matching

Action: analyze

Use the analyze_failure MCP tool from icdev-knowledge:

• error_message: from arguments
• Matches against known patterns
• Returns:
  • Matching patterns with confidence
  • Root cause determination
  • Suggested actions
  • Whether auto-healable

Action: stats

Query knowledge base statistics:

python -c "
import sqlite3
conn = sqlite3.connect('data/icdev.db')
patterns = conn.execute('SELECT COUNT(*) FROM knowledge_patterns').fetchone()[0]
heals = conn.execute('SELECT COUNT(*) FROM self_healing_events').fetchone()[0]
failures = conn.execute('SELECT COUNT(*) FROM failure_log').fetchone()[0]
print(f'Patterns: {patterns}')
print(f'Self-heal events: {heals}')
print(f'Failures logged: {failures}')
"

Self-Healing Integration

The knowledge base powers the self-healing system:

• Patterns with confidence >= 0.7 and auto_healable=true trigger automatic remediation
• Each successful heal increases the pattern's confidence and use_count
• Failed heals decrease confidence
• This creates a feedback loop that improves over time

Example

/icdev-knowledge search "database connection timeout"
/icdev-knowledge add "OOM Kill Pattern"
/icdev-knowledge recommend abc123-uuid
/icdev-knowledge analyze "ConnectionRefusedError: [Errno 111] Connection refused"
/icdev-knowledge stats

Error Handling

• If knowledge DB empty: suggest running /icdev-secure and /icdev-test to generate initial patterns
• If search returns no results: suggest broader query terms
• If pattern already exists (by name): update existing instead of duplicate