Files
lennyzeltser-conversation-r…/examples/ir-report.yaml
T
lennyzeltser c437f04e0b Initial commit
2026-01-14 17:55:56 -05:00

191 lines
6.8 KiB
YAML

# Multi-scenario demo: AI-Assisted IR Report Writing
# Demonstrates the tab system with three scenarios
#
# Based on: https://zeltser.com/good-ir-reports-with-ai
meta:
title: "AI-Assisted IR Report Writing"
description: "Using AI tools to create, review, and improve incident response documentation"
theme: chat
autoAdvance: true
scenarios:
- id: create
title: "IR Report Creation"
participants:
- id: analyst
label: "Security Analyst"
role: left
- id: ai
label: "AI Assistant"
role: right
steps:
- type: message
from: analyst
content: "I need to create an IR report for a BEC incident. The CFO received a fraudulent wire transfer request that appeared to be from the CEO."
- type: message
from: ai
content: "I can help you draft that IR report. Let me ask a few questions first to ensure we capture all the relevant details."
- type: annotation
content: "The AI uses a structured approach to gather incident details systematically."
- type: message
from: ai
content: "When was the incident discovered, and what was the initial indicator that something was wrong?"
- type: message
from: analyst
content: "Discovered yesterday at 2pm when the CFO called the CEO to confirm the transfer. The CEO had no knowledge of the request."
- type: message
from: ai
content: "Good catch by the CFO. What was the requested transfer amount and where was it supposed to go?"
- type: message
from: analyst
content: "$47,500 to an account at First National Bank, routing number ending in 4521."
- type: message
from: ai
content: "Based on what you've shared, here's a draft executive summary:"
codeBlock: |
EXECUTIVE SUMMARY
On [date], a Business Email Compromise (BEC) attempt
targeted our organization. An attacker impersonating
the CEO requested a $47,500 wire transfer from the CFO.
The attack was detected before funds were transferred
when the CFO verbally confirmed the request with the CEO.
Immediate actions: Email headers analyzed, account
security reviewed, finance team alerted to new protocols.
- type: transition
content: "The analyst continues building the full report..."
- id: review
title: "IR Report Review"
participants:
- id: analyst
label: "Security Analyst"
role: left
- id: ai
label: "AI Assistant"
role: right
steps:
- type: message
from: analyst
content: "Can you review this section of my IR report? I want to make sure it's clear for executives."
codeBlock: |
The threat actor utilized sophisticated social engineering
techniques leveraging compromised email infrastructure to
perpetrate a financial fraud attempt via BEC methodology.
- type: annotation
content: "The AI identifies jargon and passive voice that could confuse non-technical readers."
- type: message
from: ai
content: "This section could be clearer for executives. Here are specific suggestions:"
- type: message
from: ai
content: "1. 'sophisticated social engineering techniques' - What specifically made it sophisticated? Was it personalized?"
footnote: "Executives appreciate specifics over vague adjectives"
- type: message
from: ai
content: "2. 'leveraging compromised email infrastructure' - Whose email was compromised? The CEO's? An external vendor's?"
- type: message
from: ai
content: "3. 'perpetrate a financial fraud attempt via BEC methodology' - This is passive and jargon-heavy. Try: 'The attacker posed as our CEO to request a fraudulent wire transfer.'"
- type: message
from: analyst
content: "Good points. Let me revise that paragraph."
- type: message
from: analyst
content: "Revised version:"
codeBlock: |
An attacker gained access to the CEO's email account
and sent a convincing wire transfer request to the CFO.
The email matched the CEO's writing style and referenced
an actual pending acquisition, showing the attacker had
researched our company.
- type: message
from: ai
content: "Much better. This version is specific, active voice, and tells executives exactly what happened without requiring security background knowledge."
- id: coach
title: "Writing Coach"
participants:
- id: analyst
label: "Security Analyst"
role: left
- id: ai
label: "Writing Coach"
role: right
steps:
- type: message
from: analyst
content: "I struggle with executive summaries. They either end up too technical or too vague. Any tips?"
- type: message
from: ai
content: "Executive summaries for IR reports should answer four questions in plain language:"
- type: message
from: ai
content: "1. WHAT happened? (One sentence describing the incident)"
- type: message
from: ai
content: "2. WHAT's the impact? (Business terms: financial, operational, reputational)"
- type: message
from: ai
content: "3. WHAT did we do? (Immediate response actions)"
- type: message
from: ai
content: "4. WHAT's next? (Recommended follow-up actions)"
- type: annotation
content: "This framework helps analysts structure their thinking and ensures coverage of key points."
- type: message
from: analyst
content: "That's helpful. What about tone? My manager says my reports sound too alarmist."
- type: message
from: ai
content: "Good question. Compare these two approaches:"
codeBlock: |
ALARMIST:
"A catastrophic security breach nearly resulted in
massive financial losses. Urgent action is required
immediately to prevent future attacks."
MEASURED:
"A BEC attempt was detected and blocked before any
funds transferred. We recommend three process
improvements to reduce future risk."
- type: message
from: ai
content: "The measured version states facts, acknowledges what went right, and focuses on actionable improvements rather than fear."
- type: message
from: analyst
content: "I see the difference. Facts over fear, and focus on what we're doing about it."
- type: message
from: ai
content: "Exactly. Executives want to know the situation is under control and what resources you need. Alarmist language undermines confidence in the security team."