This course focuses on getting real, day-to-day value out of Claude and Cowork. Students start with the fundamentals that determine whether an AI workflow succeeds or fails—writing prompts as constraints, managing context, memory, and files, choosing between a skill and a prompt, wiring up connectors and MCPs, using Plan Mode, and putting safety rails around destructive actions. From there, the course works through a series of real-world use cases drawn from engineering, security, operations, and knowledge work—from building a custom Snowflake MCP and running an IAM audit at scale, to roasting postmortems, deploying on a comment, onboarding new hires in M365, and triaging an inbox of 247 unread messages. By the end, students will know how to design Claude workflows that are predictable, auditable, and genuinely useful in their own environment.
Overview
COURSE DIFFICULTY
COURSE DURATION
1h 35m
Skills Learned
After completing this online training course, students will be able to:
Write prompts that act as guardrails and constraints, not just instructions
Manage context, memory, and files so Claude has what it needs and nothing it shouldn’t
Decide when to use a skill vs. a one-off prompt and design each accordingly
Connect Claude to business systems using connectors and custom MCP servers
Use Plan Mode and confirmation patterns to keep destructive actions safe
Apply Claude to engineering workflows: data queries, network changes, IAM audits, postmortems, and deployments
Apply Claude to security workflows: phishing reporting, incident response, and ticket triage
Build cross-app workflows for knowledge work: M365 onboarding, QBRs, inbox triage, and customer dispatch
Critically evaluate AI output by alternating between skeptic and enthusiast perspectives
Access to Claude Cowork
01. Cowork and Claude Fundamentals
- Prompts as Constraints
- Context, Memory, Files
- Skill or Prompt?
- Connectors and MCPs
- Plan Mode
- Confirm Before Destructive
02. Technical and Engineering Use Cases
- Custom MCP for Snowflake
- Plan Mode for Network Change
- IAM Audit at Scale
- Phishing Leaderboard with Privacy
- Roast My Postmortems
- ServiceNow Skill for Tickets
- Deploy on a Comment
03. Practical Scenarios for Knowledge Workers
- Skeptic vs. Enthusiast
- M365 Onboarding Skill
- Dispatch from the Tarmac
- The Weekend Hero Hook
- Multi-Tool Incident Response
- Cross-App QBR
- Inbox of 247
- The Firm on the Stack
SKILLS LEARNED
Skills Learned
After completing this online training course, students will be able to:
Write prompts that act as guardrails and constraints, not just instructions
Manage context, memory, and files so Claude has what it needs and nothing it shouldn’t
Decide when to use a skill vs. a one-off prompt and design each accordingly
Connect Claude to business systems using connectors and custom MCP servers
Use Plan Mode and confirmation patterns to keep destructive actions safe
Apply Claude to engineering workflows: data queries, network changes, IAM audits, postmortems, and deployments
Apply Claude to security workflows: phishing reporting, incident response, and ticket triage
Build cross-app workflows for knowledge work: M365 onboarding, QBRs, inbox triage, and customer dispatch
Critically evaluate AI output by alternating between skeptic and enthusiast perspectives
PREREQUISITES
Access to Claude Cowork
COURSE OUTLINE
01. Cowork and Claude Fundamentals
- Prompts as Constraints
- Context, Memory, Files
- Skill or Prompt?
- Connectors and MCPs
- Plan Mode
- Confirm Before Destructive
02. Technical and Engineering Use Cases
- Custom MCP for Snowflake
- Plan Mode for Network Change
- IAM Audit at Scale
- Phishing Leaderboard with Privacy
- Roast My Postmortems
- ServiceNow Skill for Tickets
- Deploy on a Comment
03. Practical Scenarios for Knowledge Workers
- Skeptic vs. Enthusiast
- M365 Onboarding Skill
- Dispatch from the Tarmac
- The Weekend Hero Hook
- Multi-Tool Incident Response
- Cross-App QBR
- Inbox of 247
- The Firm on the Stack
