Kurrent

The product is a development tool that records and evaluates coding agent sessions to identify risky behavior, scope drift, quality gaps, and inefficient execution before code reaches production. It is sold to software developers and engineering teams in business environments who build software with coding agents. It is delivered as a SaaS platform with a configurable evaluation rubric across safety, plan adherence, quality, and efficiency that turns recurring mistakes into reusable guardrails.

Key features

• Session trace recording (JSONL)
• Safety evaluation
• Plan adherence evaluation
• Quality evaluation
• Efficiency evaluation
• Configurable evaluation rubric
• Session retrospectives with scores
• Fact extraction from sessions
• Fact clustering and deduplication
• Guardrail injection from facts
• Joint live sessions
• Session handoffs with context
• Session recall from past work
• Agent-driven code review
• Agent-assisted code review