Each example inDocumentation Index
Fetch the complete documentation index at: https://kensa.sh/docs/llms.txt
Use this file to discover all available pages before exploring further.
examples/ demonstrates a different agent domain with realistic stakes. Each one is a complete, runnable agent so you can point kensa at it end-to-end before pointing kensa at your own repo. Generate scenarios from a capture (kensa capture then kensa generate) or write them by hand against the documented behavior.
SQL Analyst
Domain: Finance / BIPath:
examples/sql-analystNatural-language-to-SQL against a SQLite database. Great for catching hallucinated metrics, bad joins, and incorrect aggregations.Incident Triage
Domain: SREPath:
examples/incident-triageLangGraph ReAct agent that classifies severity with a runbook knowledge base. Great for paging logic, severity routing, and escalation behavior.Code Reviewer
Domain: SecurityPath:
examples/code-reviewerReview agent that checks 16 rules across 8 categories (injection, secrets, auth, crypto, concurrency, error-handling, data, performance). Great for false-positive pressure and missed-critical-bug evaluation.Customer Support
Domain: CXPath:
examples/customer-supportTicket classification and response with policy constraints. Great for routing, policy adherence, and tone-sensitive outputs.SDR Qualifier
Domain: SalesPath:
examples/sdr-qualifierLead-scoring and outreach agent with ICP segments and disqualification rules. Great for evaluating prioritization and follow-up judgment.Pick the right example
| If you want to test | Start with |
|---|---|
| SQL correctness and tool use | examples/sql-analyst |
| Severity routing and runbook behavior | examples/incident-triage |
| Security review quality | examples/code-reviewer |
| Policy-aware support responses | examples/customer-support |
| Qualification and outreach logic | examples/sdr-qualifier |
Run any example
These examples are intentionally opinionated and high-stakes. They are meant to show where evals matter, not just provide toy demos.