feat: Interactive service auditing in batches

src/cloudwatch-appsignals-mcp-server/README.md

@@ -588,10 +588,37 @@ analyze_canary_failures(canary_name="webapp-erorrpagecanary")
 
 ## Recommended Workflows
 
-### 🎯 Primary Audit Workflow (Most Common)
+### 🎯 Primary Service Audit Workflow (Interactive Batch Processing)
+
+#### **For Large Service Lists (>10 services) - Interactive Batch Mode:**
 1. **Start with `audit_services()`** - Use wildcard patterns for automatic service discovery
-2. **Review findings summary** - Let user choose which issues to investigate further
-3. **Deep dive with `auditors="all"`** - For selected services needing root cause analysis
+   ```
+   audit_services(service_targets='[{"Type":"service","Data":{"Service":{"Type":"Service","Name":"*"}}}]')
+   ```
+2. **System automatically starts batch processing** - Processes first batch of 10 services
+3. **🚨 CRITICAL: When findings are discovered in a batch:**
+   - **STOP processing immediately**
+   - **Present complete audit findings** to user in clear summary
+   - **ALWAYS ASK USER TO CHOOSE:**
+     - **Option A:** Investigate specific finding with `auditors="all"`
+     - **Option B:** Continue processing with `continue_audit_batch(session_id)`
+   - **WAIT for user decision** - Never auto-continue when findings exist
+4. **✅ When batch has NO findings (all services healthy):**
+   - **Auto-continue** to next batch with `continue_audit_batch(session_id)`
+5. **Repeat** batch processing cycle with user choice at each step when findings exist
+6. **When all services processed** - Summarize audit results from all batches
+7. **Cleanup after completion** - Call `cleanup_audit_sessions()` to free memory resources
+
+#### **For Small Service Lists (≤10 services) - Direct Processing:**
+1. **Start with `audit_services()`** - Processes all services immediately
+2. **Present all audit results** showing summary of findings
+3. **🚨 IF FINDINGS EXIST:** Ask user which specific finding to investigate
+4. **WAIT for user decision** before performing targeted root cause analysis
+5. **Targeted investigation** - Use `auditors="all"` for user-selected finding only
+
+#### **Available Batch Management Tools:**
+- **`continue_audit_batch(session_id)`** - Continue to next batch in active session
+- **`cleanup_audit_sessions()`** - Clean up all batch sessions and free memory
 
 ### 🔍 SLO Investigation Workflow
 1. **Use `get_slo()`** - Understand SLO configuration and thresholds

src/cloudwatch-appsignals-mcp-server/awslabs/cloudwatch_appsignals_mcp_server/batch_processing_utils.py

@@ -0,0 +1,209 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Utilities for interactive batch processing of audit operations."""
+
+import json
+import uuid
+from datetime import datetime, timezone
+from loguru import logger
+from typing import Any, Dict, List, Optional
+
+
+# Global storage for batch sessions (in production, this would be in a database)
+_batch_sessions: Dict[str, Dict[str, Any]] = {}
+
+
+def create_batch_session(
+    targets: List[Dict[str, Any]],
+    input_obj: Dict[str, Any],
+    region: str,
+    banner: str,
+    batch_size: int = 10,
+    auto_complete: Optional[bool] = None,
+) -> str:
+    """Create a new batch processing session.
+
+    Args:
+        targets: List of all targets to process
+        input_obj: Base input object for API calls
+        region: AWS region
+        banner: Banner text for display
+        batch_size: Number of targets per batch
+        auto_complete: If True, process all batches automatically. If False, use interactive mode.
+                      If None, auto-decide based on target count.
+
+    Returns:
+        Session ID for tracking the batch processing
+    """
+    session_id = str(uuid.uuid4())
+
+    # Auto-decide batch processing mode if not specified
+    if auto_complete is None:
+        auto_complete = len(targets) <= batch_size  # Auto-complete for small lists
+
+    # Create batches
+    batches = []
+    for i in range(0, len(targets), batch_size):
+        batch = targets[i : i + batch_size]
+        batches.append(batch)
+
+    now = datetime.now(timezone.utc).isoformat()
+    session = {
+        'session_id': session_id,
+        'created_at': now,
+        'last_activity': now,
+        'targets': targets,
+        'input_obj': input_obj,
+        'batches': batches,
+        'current_batch_index': 0,
+        'processed_batches': [],
+        'failed_batches': [],
+        'all_findings': [],
+        'auto_complete': auto_complete,
+        'status': 'created',
+    }
+
+    _batch_sessions[session_id] = session
+    logger.info(f'Created batch session {session_id} with {len(batches)} batches')
+
+    return session_id
+
+
+def get_batch_session(session_id: str) -> Optional[Dict[str, Any]]:
+    """Get batch session by ID."""
+    return _batch_sessions.get(session_id)
+
+
+def update_batch_session_activity(session_id: str) -> None:
+    """Update last activity timestamp for session."""
+    if session_id in _batch_sessions:
+        _batch_sessions[session_id]['last_activity'] = datetime.now(timezone.utc).isoformat()
+
+
+def _create_batch_metadata(
+    session: Dict[str, Any], current_batch: List[Dict[str, Any]]
+) -> Dict[str, Any]:
+    """Create common batch metadata."""
+    current_index = session['current_batch_index']
+    return {
+        'batch_index': current_index + 1,
+        'total_batches': len(session['batches']),
+        'targets_in_batch': len(current_batch),
+        'targets': current_batch,
+    }
+
+
+def _build_api_input(
+    session: Dict[str, Any], current_batch: List[Dict[str, Any]]
+) -> Dict[str, Any]:
+    """Build API input object for the current batch."""
+    batch_input = {
+        'StartTime': datetime.fromtimestamp(session['input_obj']['StartTime'], tz=timezone.utc),
+        'EndTime': datetime.fromtimestamp(session['input_obj']['EndTime'], tz=timezone.utc),
+        'AuditTargets': current_batch,
+    }
+    if 'Auditors' in session['input_obj']:
+        batch_input['Auditors'] = session['input_obj']['Auditors']
+    return batch_input
+
+
+def _update_session_after_batch(session: Dict[str, Any], batch_result: Dict[str, Any]) -> None:
+    """Update session state after processing a batch."""
+    session['current_batch_index'] += 1
+
+    if batch_result['status'] == 'success':
+        session['processed_batches'].append(batch_result)
+        session['all_findings'].extend(batch_result.get('findings', []))
+    else:
+        session['failed_batches'].append(batch_result)
+
+    # Update overall session status
+    session['status'] = (
+        'completed' if session['current_batch_index'] >= len(session['batches']) else 'in_progress'
+    )
+
+
+def process_next_batch(session_id: str, appsignals_client) -> Dict[str, Any]:
+    """Process the next batch in the session.
+
+    Returns:
+        Dictionary with batch results and session status
+    """
+    session = get_batch_session(session_id)
+    if not session:
+        return {'error': 'Session not found or expired'}
+
+    update_batch_session_activity(session_id)
+
+    current_index = session['current_batch_index']
+    batches = session['batches']
+
+    if current_index >= len(batches):
+        return {'error': 'No more batches to process', 'status': 'completed'}
+
+    current_batch = batches[current_index]
+    batch_metadata = _create_batch_metadata(session, current_batch)
+
+    try:
+        # Build and execute API call
+        batch_input = _build_api_input(session, current_batch)
+        response = appsignals_client.list_audit_findings(**batch_input)
+
+        # Create success result
+        batch_findings = response.get('AuditFindings', [])
+        batch_result = {
+            **batch_metadata,
+            'findings_count': len(batch_findings),
+            'findings': batch_findings,
+            'status': 'success',
+        }
+
+    except Exception as e:
+        # Create error result
+        batch_result = {**batch_metadata, 'error': str(e), 'status': 'failed'}
+
+    # Update session state and return result
+    _update_session_after_batch(session, batch_result)
+    return batch_result
+
+
+def cleanup_batch_sessions() -> None:
+    """Clean up all batch sessions from memory."""
+    global _batch_sessions
+
+    initial_count = len(_batch_sessions)
+    _batch_sessions.clear()
+    logger.info(f'Cleaned up all {initial_count} batch sessions')
+
+
+def format_batch_result(batch_result: Dict[str, Any], session: Dict[str, Any]) -> str:
+    """Format batch processing result for user display with essential information only."""
+    batch_index = batch_result['batch_index']
+    total_batches = batch_result['total_batches']
+
+    if batch_result.get('error'):
+        return f'❌ Batch {batch_index}/{total_batches} failed: {batch_result["error"]}'
+
+    findings_count = len(batch_result.get('findings', []))
+
+    if findings_count == 0:
+        status = f'✅ Batch {batch_index}/{total_batches}: {batch_result["targets_in_batch"]} services healthy'
+        if batch_index < total_batches:
+            status += f" | Continue: continue_audit_batch('{session['session_id']}')"
+        return status
+
+    # Keep full JSON for MCP observation when findings exist
+    findings_json = json.dumps(batch_result['findings'], indent=2, default=str)
+    return f'⚠️ Batch {batch_index}/{total_batches}: {findings_count} findings\n```\n{findings_json}\n```'

src/cloudwatch-appsignals-mcp-server/awslabs/cloudwatch_appsignals_mcp_server/batch_tools.py

@@ -0,0 +1,84 @@
+# Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""Batch processing tools for interactive audit workflows."""
+
+from .aws_clients import appsignals_client
+from .batch_processing_utils import (
+    cleanup_batch_sessions,
+    format_batch_result,
+    get_batch_session,
+    process_next_batch,
+)
+from loguru import logger
+from pydantic import Field
+
+
+async def continue_audit_batch(
+    batch_session_id: str = Field(
+        ..., description='Session ID from previous batch processing to continue'
+    ),
+) -> str:
+    """Continue processing the next batch in an active audit session.
+
+    **INTERACTIVE BATCH PROCESSING TOOL**
+    Use this tool to continue processing the next batch of targets in an ongoing audit session.
+
+    **WHEN TO USE:**
+    - **When there are no findings from the last batch** - Services appear healthy, continue to next batch
+    - **When customer wants to continue processing next batch**
+
+    **RETURNS:**
+    - Results from the next batch with progress information
+    - Full JSON findings for MCP observation and service name extraction
+    - Continuation instructions if more batches remain
+    - Error message if session is invalid or expired
+    """
+    try:
+        batch_result = process_next_batch(batch_session_id, appsignals_client)
+        session = get_batch_session(batch_session_id)
+
+        if batch_result.get('error'):
+            return f'Error: {batch_result["error"]}'
+
+        if not session:
+            return 'Error: Session not found or expired'
+
+        # Format and return batch result
+        formatted_result = format_batch_result(batch_result, session)
+
+        return formatted_result
+
+    except Exception as e:
+        logger.error(f'Error in continue_audit_batch: {e}', exc_info=True)
+        return f'Error: {str(e)}'
+
+
+async def cleanup_audit_sessions() -> str:
+    """Clean up all batch audit sessions from memory.
+
+    **BATCH SESSION CLEANUP TOOL**
+    Use this tool to free memory by cleaning up all completed batch audit sessions.
+
+    **WHEN TO USE:**
+    - After completing a full service audit to free memory
+    - To clean up all sessions and free memory resources
+    """
+    try:
+        cleanup_batch_sessions()
+        return 'All batch sessions cleaned up successfully.'
+
+    except Exception as e:
+        logger.error(f'Error in cleanup_audit_sessions: {e}', exc_info=True)
+        return f'Error: {str(e)}'