Understand the codebase. Test the agent’s behavior.
Setsumei provides reusable codebase analysis: relationships, prioritized findings, a local scan brief and MCP context. RunQC tests selected behavior at an AI-agent endpoint you are authorized to assess. They are separate products for different engineering questions.
structure / relationships
SETSUMEI
Analyze the codebase
behavior / responses
More change puts more pressure on system understanding.
Context assembled across people and tools.
AI adds more changes to understand.
Reuse analysis and inspect the results.
Codebase understanding and agent testing serve different jobs.
Where should we look, and what could a change affect?
Test how your agent behaves before users do.
Test only endpoints you are authorized to assess. Results support human review; they do not certify an agent as safe.
Explore RunQC ↗Repeat the workflow when its inputs change.
Refresh Setsumei analysis after relevant source changes, then use reports, the browser brief or MCP for the next question. With RunQC, rerun relevant scenarios when the agent, configuration or behavior changes. Your team interprets each product’s separate results.
EVIDENCE
Discuss the codebase your team needs to understand.
Describe the engineering question and language/build context at a high level. Discuss software fit and a customer-run evaluation with local results, a scan brief and relevant MCP queries.