Security, Safety and Evaluation for Claude Applications
Making Claude applications safe and dependable: guardrails, prompt injection defenses, evaluations, testing and debugging.
CLA-205 12 lessons planned ▶ Video lessons coming soon
Offered by HSM University · Claude Certification Prep
Course content
- 1Threat modeling an LLM applicationComing
- 2Prompt injection and indirect injectionComing
- 3Guardrails and input and output filteringComing
- 4Protecting secrets and personal dataComing
- 5Permissions for tools and agentsComing
- 6Building evaluation setsComing
- 7Code-graded, model-graded and human evaluationComing
- 8Regression testing for promptsComing
- 9Debugging unexpected outputsComing
- 10Monitoring in productionComing
- 11Responsible deployment checklistsComing
- 12Lab walkthrough: an evaluation harnessComing
Requirements
Recommended background. You can start any course at any time.
Description
Making Claude applications safe and dependable: guardrails, prompt injection defenses, evaluations, testing and debugging.