Save Pattern - Store Application Knowledge

What This Skill Does

Manages the full lifecycle of application patterns in AgentDB:

• Store - Create new patterns with semantic embeddings
• Update - Replace patterns (AgentDB tracks versions)
• Deprecate - Mark patterns as outdated with replacement guidance
• Delete - Remove patterns entirely

CRITICAL: What This Skill IS and IS NOT For

USE FOR (Application Knowledge)

DO NOT USE FOR (Transient/Swarm Memory)

Patterns are PERMANENT application knowledge, not transient swarm state.

The Pattern Workflow

1. BEFORE work:  get-pattern  → Check existing patterns first
2. DURING work:  Discover new approaches
3. AFTER work:   save-pattern → Store reusable discoveries (THIS SKILL)
                 reflexion    → Evaluate if get-pattern results helped

Quick Usage

Store New Pattern

npx agentdb store-pattern \
  --type "development" \
  --domain "ndp-patterns" \
  --pattern '{
    "name": "pattern-name",
    "approach": "# Pattern Name\n\n## Context\nWhen/why to use this.\n\n## Prerequisites\n- Required setup\n\n## Steps\n1. First step\n2. Second step\n\n## Example\nConcrete usage.\n\n## Verification\nHow to confirm it worked.",
    "version": "1.0",
    "related_files": ["path/to/related/file.md"],
    "last_verified": "2025-12-22"
  }' \
  --confidence 0.9

Pattern Content Template

# Pattern Name

## Context
When/why you would use this pattern.

## Prerequisites
What must be in place before starting.

## Steps
1. First step
2. Second step
3. ...

## Example
Concrete usage example or file reference.

## Verification
How to confirm it worked.

## Related
- Related patterns or files

Pattern Categories

Update Existing Pattern

Store with the same pattern name - AgentDB handles versioning:

npx agentdb store-pattern \
  --type "development" \
  --domain "ndp-patterns" \
  --pattern '{
    "name": "add-stream",
    "approach": "# Add New Data Stream (Updated)\n\n## Steps\n1. Create config...\n2. NEW: Add retention field (required since v2.0)\n3. ...",
    "version": "2.0",
    "last_verified": "2025-12-22",
    "changelog": "Added required retention field"
  }' \
  --confidence 0.85

Deprecate Pattern

When a pattern is replaced by a better approach:

npx agentdb store-pattern \
  --type "deployment" \
  --domain "ndp-patterns" \
  --pattern '{
    "name": "manual-config-sync",
    "approach": "# DEPRECATED: Manual Config Sync\n\n## Status\nDEPRECATED as of 2025-12-22\n\n## Replacement\nUse deployment:automated-config-sync instead.\n\n## Migration\n1. Stop using manual etcdctl put commands\n2. Edit YAML files in config/streams/\n3. Run: ./deploy.sh sync\n\n## Reason\nManual sync was error-prone.",
    "status": "deprecated",
    "replacement": "automated-config-sync",
    "deprecation_date": "2025-12-22"
  }' \
  --confidence 0.0

Prefer Deprecation over Deletion - deprecated patterns serve as historical reference.

Real Examples

Architecture Pattern

npx agentdb store-pattern \
  --type "architecture" \
  --domain "ndp-patterns" \
  --pattern '{
    "name": "domain-adapter-source",
    "approach": "# Domain Adapter Pattern\n\n## Context\nAll data sources implement the Source trait for uniform handling.\n\n## Implementation\n1. Create struct implementing Source trait\n2. Implement fetch() -> Vec<TimeSeriesPoint>\n3. Implement health_check() -> HealthStatus\n\n## Example\nSee: core/src/sources/http_poll.rs",
    "tags": ["architecture", "traits", "hexagonal", "source"],
    "related_files": ["core/src/sources/http_poll.rs", "core/src/traits.rs"]
  }' \
  --confidence 0.95

Procedure Pattern

npx agentdb store-pattern \
  --type "procedures" \
  --domain "ndp-patterns" \
  --pattern '{
    "name": "add-grafana-dashboard",
    "approach": "# Add Grafana Dashboard\n\n## Prerequisites\n- DuckDB datasource configured (uid: duckdb-ndp)\n- Stream data in Parquet bronze layer\n\n## Steps\n1. Create JSON in config/grafana/dashboards/\n2. Use time_bucket() for aggregation\n3. Filter by stream_id in WHERE clause\n4. Handle unit conversions in SQL\n\n## Verification\nRestart Grafana, check dashboard loads without errors",
    "tags": ["procedures", "grafana", "dashboard", "visualization"],
    "related_files": ["config/grafana/dashboards/indoor-vs-outdoor.json"]
  }' \
  --confidence 0.9

Check Pattern Health

npx agentdb db stats

Returns: total patterns, average success rates, top task types, patterns needing review.

Best Practices

1. Be Specific - Include concrete examples, not just theory
2. Include Context - Explain when/why to use the pattern
3. Reference Files - Link to actual code that implements the pattern
4. Use Consistent Tags - Category + descriptive tags (kebab-case)
5. Deprecate Don't Delete - Preserve knowledge unless harmful
6. Set Realistic Success Rates - Start at 0.8-0.9, let feedback adjust
7. Include Verification Steps - How to confirm the pattern worked
8. Store Only Reusable Knowledge - Not one-off solutions

Related Skills

• get-pattern - Retrieve patterns BEFORE work (always check first)
• reflexion - Record feedback on whether get-pattern results helped

NOT related to:

• Swarm coordination (use claude-flow tools)
• Transient task memory (use claude-flow memory with TTL)
• Agent state management (use claude-flow tools)