Reference Lookup

Overview

Before starting any non-trivial implementation, search your knowledge base. You may have already studied a repo that solves this exact problem — or at least has relevant patterns you can adapt.

When to Use

• Before implementing a new feature or module
• When choosing between approaches (check what worked in references)
• When debugging a class of problem you've seen before
• When evaluating a new library or technique

Don't Skip This Step

"I'll figure it out from scratch" costs 2 hours. "Let me check my references first" costs 5 minutes and often saves the 2 hours.

---

Lookup Workflow

Step 1 — Define What You're Looking For

Write a one-sentence description of your task, then extract:

Task: "Implement incremental Delta Lake writes that avoid full overwrites"

Keywords: Delta, incremental, merge, upsert, MERGE INTO, idempotent
Domain:   databricks, patterns
Tech:     PySpark, Delta Lake

Step 2 — Search Context KB

# Search by keyword across all context
grep -r "merge\|upsert\|incremental" ~/.copilot/context/ -l --include="*.md"

# Read matching files
cat ~/.copilot/context/databricks/{matching-file}.md

Step 3 — Search Reference Analyses

# Search pattern sections of analysis files
grep -r "incremental\|merge\|delta" ~/references/ --include="analysis.md" -l
grep -r "incremental" ~/references/ --include="analysis.md" -n | head -20

Step 4 — Browse Reference Code

If an analysis mentions a specific file:

grep -r "MERGE INTO\|delta.merge" ~/references/{repo}/ --include="*.py" -n | head -20
cat ~/references/{repo}/{path/to/file}.py

---

Presenting Results

After searching, produce a structured summary:

## Lookup Results: "{your task}"

### ✅ Direct Match
Found in: ~/.copilot/context/databricks/delta-merge-pattern.md
---
{paste key content / snippet}
---
→ Apply by: {specific adaptation needed}

### 🔶 Related
- ~/references/delta-lake-recipes/analysis.md §6 — "MERGE pattern with SCD Type 2"
  Worth reading: covers deduplication during merge

### ❌ Not Found
No existing pattern for "{specific aspect of task}"
→ After implementing, consider adding to ~/.copilot/context/patterns/

### Recommended Approach
Based on references, use: {specific recommendation with reasoning}

---

Search Cheatsheet

Looking for	Search command
All entries in a domain	ls ~/.copilot/context/{domain}/
Keyword across all KB	grep -r "{term}" ~/.copilot/context/ -l
Keyword in references	grep -r "{term}" ~/references/ --include="*.md" -l
Code snippet in refs	grep -r "{term}" ~/references/ --include="*.py" -n
What repos cover topic	grep -r "{term}" ~/references/ --include="analysis.md" -l

---

After Lookup

Result	Action
Direct match found	Use it, note any adaptation needed
Partial match found	Build on it, document the gap
No match found	Implement, then run context-extraction to capture it
Relevant repo not analyzed	Run repo-analysis skill first

Tips

• Cast a wide net with keywords — don't just search the exact term
• Read, don't skim — a 2-minute read of a matching file is worth it
• Check multiple domains — a pattern from auth/ might apply to patterns/ use cases
• Use reference-navigator agent for an automated lookup with AI synthesis