Personality evals for LLMs.
Define target personas, run batch evals on checkpoints, monitor production responses. Open source.
Launching Lindr Cloud Soon
Join the waitlist to get early access.
Try It Now
Paste any text from your chatbot or AI assistant. See its personality profile across 10 dimensions in seconds.
No analysis yet
Paste text and click Analyze to see results
Open Source. Self-Host Today.
Don't want to wait? Deploy Lindr on your own infrastructure. Full functionality, no limits, completely free.
View on GitHubBuilt for AI Engineers
Most observability tools measure latency and errors. Lindr measures behavioral consistency—whether your agent acts the way you designed it to.
Multi-dimensional Evals
10-dimension personality vectors based on psychometric research. Compare model checkpoints, detect regressions, and validate fine-tuning.
persona.diff(baseline, checkpoint)Real-time Monitoring
Drop-in middleware for your LLM gateway. Async analysis, configurable alerting thresholds, zero latency impact on user responses.
lindr.monitor(response, persona)Drift Detection
Statistical analysis across evaluation windows. Catch behavioral drift before it compounds—whether from prompt changes, model updates, or edge cases.
drift.detect(window='7d')Frontier Model Personality Benchmark
4,368 evaluations across GPT-5.2 and Claude Opus 4.5 reveal distinct personality profiles with statistically significant differences.
Claude Opus 4.5
Anthropic
GPT-5.2
OpenAI