Conversational Q&A Engine (v1.5.0)

DevDiff v1.5.0 introduces the Conversational Q&A Engine (ConversationalQA), providing instant sub-50ms codebase queries with pronoun resolution (it, this, that, them, those) and multi-turn context memory across session turns.

---

🎯 Architecture & Performance

---

⚡ Performance Breakdown

Query Type	Typical Response Time	Data Source
Entity Lookup ("What does X do?")	< 10ms	Codebase Index
Dependency Query ("What depends on X?")	< 15ms	Knowledge Graph
Count Queries ("How many functions?")	< 5ms	AST Statistics
Compliance Checks ("Is this GDPR safe?")	< 20ms	Privacy Rules Engine
Context Resolution ("Tell me about it")	< 5ms	Conversation State
AI Fallback ("Explain complex logic")	500ms - 2s	Ollama / Cloud LLM

---

💬 Question Patterns & Pronoun Resolution

The Q&A engine dynamically resolves pronouns based on conversation history:

1. "What does X do?": Looks up X in indexed classes, interfaces, components, and functions. Sets X as the active topic.

2. "What depends on it?" / "What about it?": Automatically substitutes "it" with the active topic (X).

3. "Where is X?": Returns the relative file path and line number location.

4. "How many functions / classes / files?": Returns exact numeric totals from the indexed codebase snapshot.

5. "Is this GDPR / HIPAA compliant?": Executes quick compliance checks for hardcoded secrets, unencrypted localStorage usage, and PII patterns.

---

💻 CLI Usage & Examples

# Ask initial question
devdiff ask "What does CountifyStorage do?"

# Output:
# [lucide:message-square] DevDiff Conversational Memory (12ms from index):
# CountifyStorage in `js/storage.js` — IndexedDB wrapper providing async CRUD.
# [lucide:file-text] Sources: `js/storage.js`
# [lucide:lightbulb] Follow up:
#    • "What depends on CountifyStorage?"
#    • "Show me the code"

# Natural follow-up (uses pronoun context):
devdiff ask "What depends on it?"

# Output:
# CountifyStorage is used by 2 files: `js/app.js` and `js/analytics.js`.

---

🔧 Programmatic API (@eldrex/core)

import { ConversationalQA } from "@eldrex/core";

const qa = new ConversationalQA(process.cwd());

// Turn 1
const res1 = await qa.ask("What does SkillManager do?");
console.log(res1.answer); // "SkillManager in packages/core/src/skill/skill-manager.ts..."

// Turn 2 (Pronoun resolution)
const res2 = await qa.ask("What depends on it?");
console.log(res2.answer); // "SkillManager is referenced in packages/cli/src/commands/skill.ts..."

// Clear conversation context
qa.reset();