dbagent-cli 0.2.2__tar.gz → 0.3.1__tar.gz

dbagent_cli-0.3.1/PKG-INFO

@@ -0,0 +1,198 @@
+Metadata-Version: 2.4
+Name: dbagent-cli
+Version: 0.3.1
+Summary: Enterprise AI Database Copilot with Long-Term Change Recovery (CLI)
+Author: Santhosh Gupta
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+Requires-Dist: typer>=0.9.0
+Requires-Dist: rich>=13.0.0
+Requires-Dist: prompt-toolkit>=3.0.0
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: sqlalchemy>=2.0.0
+Requires-Dist: requests>=2.28.0
+Requires-Dist: httpx>=0.24.0
+Requires-Dist: python-dotenv>=1.0.0
+Provides-Extra: all
+Requires-Dist: psycopg2-binary>=2.9.0; extra == "all"
+Requires-Dist: pymysql>=1.0.0; extra == "all"
+Requires-Dist: pymongo>=4.0.0; extra == "all"
+Requires-Dist: duckdb>=0.9.0; extra == "all"
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0.0; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.20.0; extra == "dev"
+
+# ⚡ DB-Agent
+
+> **Enterprise AI Database Copilot & Autonomous Data Agent (CLI)**  
+> *Author: Santhosh Gupta | 100% Free, Standalone, Works Offline (Ollama) or with Cloud AI (Gemini / Groq / OpenRouter).*
+
+[![PyPI Version](https://img.shields.io/pypi/v/dbagent-cli.svg)](https://pypi.org/project/dbagent-cli/)
+[![Python Version](https://img.shields.io/pypi/pyversions/dbagent-cli.svg)](https://pypi.org/project/dbagent-cli/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+---
+
+## 🌟 What is DB-Agent?
+
+**DB-Agent** is a production-grade AI Database Agent and Copilot. It empowers developers, DBAs, data analysts, and non-technical users to inspect schemas, discover complex multi-hop table relationships, generate and optimize SQL, execute queries with enterprise safety guardrails, and visualize data directly in the terminal.
+
+---
+
+## 🚀 2-Minute Quickstart
+
+### Step 1: Install from PyPI
+```bash
+pip install "dbagent-cli[all]"
+```
+
+### Step 2: 1-Click AI Setup
+```bash
+db-agent setup
+```
+Choose between:
+1. **Local Offline AI (Ollama)**: Zero API keys, 100% private, runs entirely on your machine.
+2. **Free Cloud AI (Gemini / Groq / OpenRouter)**: Ultra-fast cloud processing.
+3. **Database Tools Only**: Standard DB introspection and CLI tools without AI.
+
+### Step 3: Connect Database & Save Alias
+```bash
+db-agent connect postgresql://user:password@localhost:5432/mydb --alias my_db
+```
+*(Supports PostgreSQL, MySQL, SQLite, SQL Server, Oracle, DuckDB, and MongoDB).*
+
+### Step 4: Ask Plain English Questions or Launch Chat!
+```bash
+# 1. Ask a question and get instant live results table + ASCII charts
+db-agent ask "list top 10 users" --db my_db
+
+# 2. Or start the interactive AI copilot shell
+db-agent chat --db my_db
+```
+
+---
+
+## 🛡️ Key Enterprise Features
+
+### 1. 🧠 Relationship Graph & Smart Multi-Hop JOINs
+Automatically discovers multi-hop join paths between disconnected tables without guessing arbitrary joins.
+```bash
+# Find shortest relationship path between two tables:
+db-agent path users events --db my_db
+
+# Explain join cardinality (1:1, 1:N) and recommended join syntax:
+db-agent explain-join customers orders --db my_db
+```
+
+### 2. ⚡ Persistent Schema Cache & Change Detection
+Never runs expensive full database scans on every query. Local cache (`~/.dbagent/cache/`) detects schema diffs:
+```bash
+db-agent refresh-schema --db my_db
+db-agent schema-status --db my_db
+```
+
+### 3. 🔒 Safety Modes & Production Protection
+Five operational safety modes enforce strict boundaries before SQL execution:
+* **`SAFE`**: Only `SELECT`, `SHOW`, `DESCRIBE`, `EXPLAIN`, `WITH` allowed. All writes and DDL are strictly blocked.
+* **`READ_ONLY`**: Enforced on `PRODUCTION` environments with 5s timeouts and row caps.
+* **`CONFIRM`**: Reads auto-execute; write and DDL operations require interactive confirmation.
+* **`SANDBOX`**: Writes executed inside uncommitted transactions with before/after diff preview.
+* **`ADMIN`**: Full administrative access with audit tracking.
+
+### 4. ⏪ Change Sets & Snapshot Rollback
+Before executing modifying queries, DB-Agent captures a recovery snapshot and creates an auditable ChangeSet:
+```bash
+db-agent changes
+db-agent changes-show CS-20260828-001
+db-agent rollback CS-20260828-001 --db my_db
+```
+
+### 5. 🏥 Database Doctor & Query Optimizer
+Run comprehensive health checks on connectivity, latency, unindexed foreign keys, and slow queries:
+```bash
+# Database health diagnostics:
+db-agent doctor --db my_db
+
+# EXPLAIN plan analysis & index recommendations:
+db-agent optimize "SELECT * FROM events WHERE camera_id = 42" --db my_db
+```
+
+### 6. 📚 Business Semantic Dictionary & Query Templates
+Teach DB-Agent your domain-specific metrics and definitions:
+```bash
+# Define a business metric:
+db-agent define offline_camera "A camera with no heartbeat for > 2 minutes" --sql "last_heartbeat < NOW() - INTERVAL '2 mins'"
+
+# List all definitions:
+db-agent definitions
+
+# Manage trusted SQL templates:
+db-agent template list
+```
+
+### 7. 🎭 Sensitive Data Masking & RBAC
+Automatically classifies columns into PII, Secrets, and Financial data, masking confidential values in outputs (`s***@gmail.com`, `**********`, `****-****-****-1234`).
+
+### 8. 📊 Automatic Visualization Detector
+Detects when query results represent time series, category distributions, or single metrics, and renders ASCII/Unicode charts directly in your terminal.
+
+---
+
+## 📖 Complete CLI Command Reference
+
+| Command | How to Use | Purpose |
+|---|---|---|
+| **`ask`** | `db-agent ask "list top 10 users" --db my_db` | Natural language to SQL -> executes & renders table. |
+| **`chat`** | `db-agent chat --db my_db` | Interactive terminal copilot with multi-turn context. |
+| **`doctor`** | `db-agent doctor --db my_db` | Comprehensive database and AI health diagnostic check. |
+| **`path`** | `db-agent path users events --db my_db` | Shortest multi-hop foreign key join path discovery. |
+| **`explain-join`** | `db-agent explain-join customers orders` | Explains join relationship, cardinality, and syntax. |
+| **`optimize`** | `db-agent optimize "<SQL>" --db my_db` | Analyzes EXPLAIN plan and recommends indexes. |
+| **`scan`** | `db-agent scan --db my_db --export db.md` | Full database schema introspection & Markdown export. |
+| **`refresh-schema`**| `db-agent refresh-schema --db my_db` | Refreshes schema cache and reports detected diffs. |
+| **`schema-status`** | `db-agent schema-status --db my_db` | Displays schema cache age, version, and health. |
+| **`changes`** | `db-agent changes` | Lists recorded write ChangeSets. |
+| **`changes-show`**| `db-agent changes-show CS-20260828-001` | Shows before/after snapshot of a ChangeSet. |
+| **`rollback`** | `db-agent rollback CS-20260828-001` | Generates and executes inverse rollback SQL. |
+| **`define`** | `db-agent define revenue "SUM(order.amount)"` | Saves a business definition into the semantic layer. |
+| **`definitions`** | `db-agent definitions` | Lists all defined business terminology. |
+| **`template`** | `db-agent template list` | Manages reusable, parameterized query templates. |
+| **`audit`** | `db-agent audit` | Displays recent query execution audit log. |
+| **`connect`** | `db-agent connect <url> --alias my_db` | Tests connection and saves profile alias. |
+| **`use`** | `db-agent use gemini` | Switches default AI provider. |
+| **`run`** | `db-agent run "SELECT * FROM users" --db my_db` | Executes raw SQL query with Rich table output. |
+| **`generate`** | `db-agent generate "ETL script" --db my_db -o etl.py` | Generates Python ETL, Alembic migrations, or APIs. |
+| **`setup`** | `db-agent setup` | 1-Click setup wizard for local or cloud AI. |
+| **`config`** | `db-agent config` | Interactive AI API keys and preferences wizard. |
+| **`profiles`** | `db-agent profiles` | Lists all saved database connection profiles. |
+| **`models`** | `db-agent models` | Status of local Ollama models and cloud AI providers. |
+| **`guide`** | `db-agent guide` | Displays full interactive command cheatsheet. |
+
+---
+
+## 💬 In-Chat REPL Commands (`db-agent chat`)
+
+Inside the interactive chat shell, you can use built-in colon commands:
+
+* **`:doctor`** — Run database health diagnostics.
+* **`:path <t1> <t2>`** — Discover shortest join path between tables.
+* **`:join <t1> <t2>`** — Explain join cardinality and recommended join syntax.
+* **`:mode <SAFE|CONFIRM|SANDBOX|ADMIN>`** — Switch safety mode mid-session.
+* **`:changes`** — List recent write ChangeSets.
+* **`:tables`** — List all database tables.
+* **`:table <name>`** — Inspect columns, types, primary keys, and sample data.
+* **`:fk <name>`** — View incoming and outgoing foreign keys.
+* **`:provider <name>`** — Switch AI provider (`ollama`, `gemini`, `groq`, `openrouter`).
+* **`:model <name>`** — Switch active model name on the fly.
+* **`:explain`** — Show generated SQL and query analysis for the last turn.
+* **`:history`** — View query history for the session.
+* **`:auto` / `:noauto`** — Toggle automatic execution of read queries.
+* **`:run`** — Re-execute the last generated SQL query.
+* **`:clear`** — Clear multi-turn conversational context.
+* **`:export <file>`** — Save the last generated SQL to a file.
+* **`:exit`** — Exit chat session.
+
+---
+
+## 📄 License
+MIT License. Built with ❤️ by Santhosh Gupta.

dbagent_cli-0.3.1/README.md

@@ -0,0 +1,174 @@
+# ⚡ DB-Agent
+
+> **Enterprise AI Database Copilot & Autonomous Data Agent (CLI)**  
+> *Author: Santhosh Gupta | 100% Free, Standalone, Works Offline (Ollama) or with Cloud AI (Gemini / Groq / OpenRouter).*
+
+[![PyPI Version](https://img.shields.io/pypi/v/dbagent-cli.svg)](https://pypi.org/project/dbagent-cli/)
+[![Python Version](https://img.shields.io/pypi/pyversions/dbagent-cli.svg)](https://pypi.org/project/dbagent-cli/)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+
+---
+
+## 🌟 What is DB-Agent?
+
+**DB-Agent** is a production-grade AI Database Agent and Copilot. It empowers developers, DBAs, data analysts, and non-technical users to inspect schemas, discover complex multi-hop table relationships, generate and optimize SQL, execute queries with enterprise safety guardrails, and visualize data directly in the terminal.
+
+---
+
+## 🚀 2-Minute Quickstart
+
+### Step 1: Install from PyPI
+```bash
+pip install "dbagent-cli[all]"
+```
+
+### Step 2: 1-Click AI Setup
+```bash
+db-agent setup
+```
+Choose between:
+1. **Local Offline AI (Ollama)**: Zero API keys, 100% private, runs entirely on your machine.
+2. **Free Cloud AI (Gemini / Groq / OpenRouter)**: Ultra-fast cloud processing.
+3. **Database Tools Only**: Standard DB introspection and CLI tools without AI.
+
+### Step 3: Connect Database & Save Alias
+```bash
+db-agent connect postgresql://user:password@localhost:5432/mydb --alias my_db
+```
+*(Supports PostgreSQL, MySQL, SQLite, SQL Server, Oracle, DuckDB, and MongoDB).*
+
+### Step 4: Ask Plain English Questions or Launch Chat!
+```bash
+# 1. Ask a question and get instant live results table + ASCII charts
+db-agent ask "list top 10 users" --db my_db
+
+# 2. Or start the interactive AI copilot shell
+db-agent chat --db my_db
+```
+
+---
+
+## 🛡️ Key Enterprise Features
+
+### 1. 🧠 Relationship Graph & Smart Multi-Hop JOINs
+Automatically discovers multi-hop join paths between disconnected tables without guessing arbitrary joins.
+```bash
+# Find shortest relationship path between two tables:
+db-agent path users events --db my_db
+
+# Explain join cardinality (1:1, 1:N) and recommended join syntax:
+db-agent explain-join customers orders --db my_db
+```
+
+### 2. ⚡ Persistent Schema Cache & Change Detection
+Never runs expensive full database scans on every query. Local cache (`~/.dbagent/cache/`) detects schema diffs:
+```bash
+db-agent refresh-schema --db my_db
+db-agent schema-status --db my_db
+```
+
+### 3. 🔒 Safety Modes & Production Protection
+Five operational safety modes enforce strict boundaries before SQL execution:
+* **`SAFE`**: Only `SELECT`, `SHOW`, `DESCRIBE`, `EXPLAIN`, `WITH` allowed. All writes and DDL are strictly blocked.
+* **`READ_ONLY`**: Enforced on `PRODUCTION` environments with 5s timeouts and row caps.
+* **`CONFIRM`**: Reads auto-execute; write and DDL operations require interactive confirmation.
+* **`SANDBOX`**: Writes executed inside uncommitted transactions with before/after diff preview.
+* **`ADMIN`**: Full administrative access with audit tracking.
+
+### 4. ⏪ Change Sets & Snapshot Rollback
+Before executing modifying queries, DB-Agent captures a recovery snapshot and creates an auditable ChangeSet:
+```bash
+db-agent changes
+db-agent changes-show CS-20260828-001
+db-agent rollback CS-20260828-001 --db my_db
+```
+
+### 5. 🏥 Database Doctor & Query Optimizer
+Run comprehensive health checks on connectivity, latency, unindexed foreign keys, and slow queries:
+```bash
+# Database health diagnostics:
+db-agent doctor --db my_db
+
+# EXPLAIN plan analysis & index recommendations:
+db-agent optimize "SELECT * FROM events WHERE camera_id = 42" --db my_db
+```
+
+### 6. 📚 Business Semantic Dictionary & Query Templates
+Teach DB-Agent your domain-specific metrics and definitions:
+```bash
+# Define a business metric:
+db-agent define offline_camera "A camera with no heartbeat for > 2 minutes" --sql "last_heartbeat < NOW() - INTERVAL '2 mins'"
+
+# List all definitions:
+db-agent definitions
+
+# Manage trusted SQL templates:
+db-agent template list
+```
+
+### 7. 🎭 Sensitive Data Masking & RBAC
+Automatically classifies columns into PII, Secrets, and Financial data, masking confidential values in outputs (`s***@gmail.com`, `**********`, `****-****-****-1234`).
+
+### 8. 📊 Automatic Visualization Detector
+Detects when query results represent time series, category distributions, or single metrics, and renders ASCII/Unicode charts directly in your terminal.
+
+---
+
+## 📖 Complete CLI Command Reference
+
+| Command | How to Use | Purpose |
+|---|---|---|
+| **`ask`** | `db-agent ask "list top 10 users" --db my_db` | Natural language to SQL -> executes & renders table. |
+| **`chat`** | `db-agent chat --db my_db` | Interactive terminal copilot with multi-turn context. |
+| **`doctor`** | `db-agent doctor --db my_db` | Comprehensive database and AI health diagnostic check. |
+| **`path`** | `db-agent path users events --db my_db` | Shortest multi-hop foreign key join path discovery. |
+| **`explain-join`** | `db-agent explain-join customers orders` | Explains join relationship, cardinality, and syntax. |
+| **`optimize`** | `db-agent optimize "<SQL>" --db my_db` | Analyzes EXPLAIN plan and recommends indexes. |
+| **`scan`** | `db-agent scan --db my_db --export db.md` | Full database schema introspection & Markdown export. |
+| **`refresh-schema`**| `db-agent refresh-schema --db my_db` | Refreshes schema cache and reports detected diffs. |
+| **`schema-status`** | `db-agent schema-status --db my_db` | Displays schema cache age, version, and health. |
+| **`changes`** | `db-agent changes` | Lists recorded write ChangeSets. |
+| **`changes-show`**| `db-agent changes-show CS-20260828-001` | Shows before/after snapshot of a ChangeSet. |
+| **`rollback`** | `db-agent rollback CS-20260828-001` | Generates and executes inverse rollback SQL. |
+| **`define`** | `db-agent define revenue "SUM(order.amount)"` | Saves a business definition into the semantic layer. |
+| **`definitions`** | `db-agent definitions` | Lists all defined business terminology. |
+| **`template`** | `db-agent template list` | Manages reusable, parameterized query templates. |
+| **`audit`** | `db-agent audit` | Displays recent query execution audit log. |
+| **`connect`** | `db-agent connect <url> --alias my_db` | Tests connection and saves profile alias. |
+| **`use`** | `db-agent use gemini` | Switches default AI provider. |
+| **`run`** | `db-agent run "SELECT * FROM users" --db my_db` | Executes raw SQL query with Rich table output. |
+| **`generate`** | `db-agent generate "ETL script" --db my_db -o etl.py` | Generates Python ETL, Alembic migrations, or APIs. |
+| **`setup`** | `db-agent setup` | 1-Click setup wizard for local or cloud AI. |
+| **`config`** | `db-agent config` | Interactive AI API keys and preferences wizard. |
+| **`profiles`** | `db-agent profiles` | Lists all saved database connection profiles. |
+| **`models`** | `db-agent models` | Status of local Ollama models and cloud AI providers. |
+| **`guide`** | `db-agent guide` | Displays full interactive command cheatsheet. |
+
+---
+
+## 💬 In-Chat REPL Commands (`db-agent chat`)
+
+Inside the interactive chat shell, you can use built-in colon commands:
+
+* **`:doctor`** — Run database health diagnostics.
+* **`:path <t1> <t2>`** — Discover shortest join path between tables.
+* **`:join <t1> <t2>`** — Explain join cardinality and recommended join syntax.
+* **`:mode <SAFE|CONFIRM|SANDBOX|ADMIN>`** — Switch safety mode mid-session.
+* **`:changes`** — List recent write ChangeSets.
+* **`:tables`** — List all database tables.
+* **`:table <name>`** — Inspect columns, types, primary keys, and sample data.
+* **`:fk <name>`** — View incoming and outgoing foreign keys.
+* **`:provider <name>`** — Switch AI provider (`ollama`, `gemini`, `groq`, `openrouter`).
+* **`:model <name>`** — Switch active model name on the fly.
+* **`:explain`** — Show generated SQL and query analysis for the last turn.
+* **`:history`** — View query history for the session.
+* **`:auto` / `:noauto`** — Toggle automatic execution of read queries.
+* **`:run`** — Re-execute the last generated SQL query.
+* **`:clear`** — Clear multi-turn conversational context.
+* **`:export <file>`** — Save the last generated SQL to a file.
+* **`:exit`** — Exit chat session.
+
+---
+
+## 📄 License
+MIT License. Built with ❤️ by Santhosh Gupta.

{dbagent_cli-0.2.2 → dbagent_cli-0.3.1}/dbagent/__init__.py

@@ -2,5 +2,5 @@
 DB-Agent: Universal Database Introspection and Script Generation AI Agent (CLI).
 """
 
-__version__ = "0.2.2"
+__version__ = "0.3.1"
 __author__ = "Santhosh Gupta"

dbagent_cli-0.3.1/dbagent/agent/confidence.py

@@ -0,0 +1,86 @@
+"""
+AI Confidence Scoring Engine.
+Computes confidence scores (0-100%) for generated SQL queries based on:
+- Table name matching precision (exact vs fuzzy vs fallback)
+- Foreign key relationship verification (verified FK path vs guessed join)
+- Ambiguity indicators and business dictionary definitions
+"""
+
+from typing import Dict, Any, List, Optional
+from dataclasses import dataclass, field
+
+
+@dataclass
+class ConfidenceScore:
+    """Confidence evaluation of an AI-generated query."""
+    score_percentage: int
+    level: str  # HIGH (85-100%), MEDIUM (60-84%), LOW (<60%)
+    rationales: List[str] = field(default_factory=list)
+    caveats: List[str] = field(default_factory=list)
+
+    def format_summary(self) -> str:
+        color = "green" if self.score_percentage >= 85 else ("yellow" if self.score_percentage >= 60 else "red")
+        lines = [f"[{color}]Confidence: {self.score_percentage}% ({self.level})[/{color}]"]
+        for r in self.rationales:
+            lines.append(f"  [dim]✓ {r}[/dim]")
+        for c in self.caveats:
+            lines.append(f"  [yellow]⚠ {c}[/yellow]")
+        return "\n".join(lines)
+
+
+class ConfidenceEvaluator:
+    """
+    Evaluates confidence score for a generated SQL statement.
+    """
+
+    @classmethod
+    def evaluate(
+        self,
+        user_prompt: str,
+        sql: str,
+        exact_tables: List[str],
+        fuzzy_tables: List[str],
+        used_known_fk: bool = True,
+        has_business_definition: bool = False,
+        retries: int = 0,
+    ) -> ConfidenceScore:
+        score = 100
+        rationales = []
+        caveats = []
+
+        # Table matching precision
+        if exact_tables:
+            rationales.append(f"Exact table match verified: {', '.join(exact_tables)}")
+        elif fuzzy_tables:
+            score -= 15
+            caveats.append(f"Fuzzy table name match: {', '.join(fuzzy_tables)}")
+        else:
+            score -= 30
+            caveats.append("No direct table name mentioned; inferred from schema keywords")
+
+        # Join relationship certainty
+        if "JOIN" in sql.upper():
+            if used_known_fk:
+                rationales.append("Verified foreign key relationship path used for JOINs")
+            else:
+                score -= 20
+                caveats.append("JOIN condition inferred without explicit foreign key constraint")
+
+        # Business definition
+        if has_business_definition:
+            rationales.append("Applied explicit business dictionary definition")
+
+        # Retries penalty
+        if retries > 0:
+            score -= (retries * 10)
+            caveats.append(f"Required {retries} automatic correction retry(ies)")
+
+        score = max(20, min(100, score))
+        level = "HIGH" if score >= 85 else ("MEDIUM" if score >= 60 else "LOW")
+
+        return ConfidenceScore(
+            score_percentage=score,
+            level=level,
+            rationales=rationales,
+            caveats=caveats,
+        )

dbagent_cli-0.3.1/dbagent/agent/conversation.py

@@ -0,0 +1,82 @@
+"""
+Multi-Turn Conversational Context Tracker.
+Maintains active tables, filter predicates, entities, and sort orders across chat turns
+so users can refine queries naturally (e.g. 'Which sites?', 'Only Hyderabad', 'Top 10?').
+"""
+
+from typing import Dict, Any, List, Optional, Set
+from dataclasses import dataclass, field
+
+
+@dataclass
+class ConversationTurn:
+    """A single turn in the conversation."""
+    user_prompt: str
+    sql: str
+    tables: List[str] = field(default_factory=list)
+    filters: List[str] = field(default_factory=list)
+    entities: Dict[str, str] = field(default_factory=dict)
+    limit: Optional[int] = None
+    is_chat: bool = False
+
+
+class ConversationContext:
+    """
+    Tracks state across interactive chat turns.
+    """
+
+    def __init__(self, max_history: int = 10):
+        self.max_history = max_history
+        self.turns: List[ConversationTurn] = []
+        self.active_tables: Set[str] = set()
+        self.active_filters: List[str] = []
+
+    def add_turn(
+        self,
+        user_prompt: str,
+        sql: str,
+        tables: Optional[List[str]] = None,
+        filters: Optional[List[str]] = None,
+        is_chat: bool = False,
+    ) -> None:
+        """Records a completed turn and updates active context."""
+        turn = ConversationTurn(
+            user_prompt=user_prompt,
+            sql=sql,
+            tables=tables or [],
+            filters=filters or [],
+            is_chat=is_chat,
+        )
+        self.turns.append(turn)
+        if len(self.turns) > self.max_history:
+            self.turns.pop(0)
+
+        if tables and not is_chat:
+            self.active_tables.update(t.lower() for t in tables)
+
+    def get_last_turn(self) -> Optional[ConversationTurn]:
+        if self.turns:
+            return self.turns[-1]
+        return None
+
+    def format_context_for_prompt(self) -> str:
+        """Formats recent context for inclusion in LLM prompt."""
+        if not self.turns:
+            return ""
+
+        recent_valid = [t for t in self.turns[-3:] if t.sql and not t.is_chat]
+        if not recent_valid:
+            return ""
+
+        lines = ["Previous Conversation Context (Use for follow-up refinements):"]
+        for i, t in enumerate(recent_valid, 1):
+            lines.append(f"  {i}. User: \"{t.user_prompt}\"")
+            lines.append(f"     Last SQL: `{t.sql.strip()}`")
+
+        return "\n".join(lines)
+
+    def clear(self) -> None:
+        """Resets the conversation history."""
+        self.turns.clear()
+        self.active_tables.clear()
+        self.active_filters.clear()