knowledge

Knowledge Recall Skill

You have access to a knowledge base of stored insights, decisions, and research findings. When a query relates to stored knowledge, relevant items are automatically injected into your context.

How It Works

Knowledge items are stored in ~/.sage/knowledge/ with keyword triggers. When your query matches those triggers, the relevant knowledge is automatically recalled and injected.

You'll see a notification like:

📚 Knowledge recalled (2)
   ├─ gdpr-summary (~450 tokens)
   └─ consent-patterns (~280 tokens)

What Gets Stored as Knowledge

Knowledge items capture durable insights that are useful across sessions:

Storage Structure

~/.sage/knowledge/
├── index.yaml              # Registry with triggers
├── global/                 # Available to all skills
│   └── api-patterns.md
└── skills/                 # Skill-scoped knowledge
    └── privacy/
        └── gdpr-summary.md

Knowledge Item Format

Each item has:

• id: Unique identifier
• keywords: Trigger words for matching
• scope: Which skills can access it (empty = all)
• content: The actual knowledge (markdown)
• source: Where it came from (optional)

Adding Knowledge

Via CLI

sage knowledge add notes.md --id gdpr-summary --keywords "gdpr,privacy,consent"
sage knowledge add api-guide.md --id api-patterns --keywords "api,rest" --skill web-dev

Via Conversation

Say "save this as knowledge" or "remember this" after discovering something worth preserving:

You: That's really useful about GDPR consent. Remember this.
Claude: [Extracts key insight and saves to knowledge base]
        Saved knowledge: gdpr-consent-requirements
        Keywords: gdpr, consent, legal basis

Querying Knowledge

Automatic

Just ask questions. If your query matches stored knowledge, it's injected automatically.

Manual

You: What do we know about GDPR?
Claude: [Recalls and presents relevant stored knowledge]

Test Matching

sage knowledge match "How does GDPR affect our API?"

Behavior

• Automatic: Knowledge is recalled based on query keywords—no explicit request needed
• Additive: Multiple items can be recalled if relevant (up to token limit)
• Scoped: Skill-specific knowledge only appears when using that skill
• Transparent: Always shows what was recalled and token cost