Examforge
A self-hostable coding assessment platform for universities and high-volume hiring that combines controlled execution, rubric scoring, candidate exports, and explicit no-AI, limited-AI, or AI-native exam policies.
Coding assessment incumbents validate a large market but package concurrency, attempts, and advanced controls for enterprise buyers. Open editor and execution components make a self-hostable alternative feasible. The source's AI-cheating detector is the dangerous part: paste patterns and tool telemetry do not prove cheating and can disadvantage legitimate workflows or accommodations. Examforge should declare the permitted tool mode before an assessment, collect only authorized evidence, use signals to route review rather than convict, and give candidates disclosure and appeal.
The technical hiring, engineering education, or assessment operations lead at an organization running repeated coding exams or large candidate cohorts.
Widespread agentic coding use forces assessment redesign now.
Hiring and university assessment operators have a concrete high-volume workflow.
The input has several graph links but moderate grounded convergence.
A clear assessment buyer, confirmed incumbent packaging, open execution and editor components, and an unserved open-core pricing shape support the direction.
The proposed prices and concurrency are unverified, cheating detection is error-prone, secure code execution is demanding, and the differentiation is partly packaging.
Discussion
No comments yet — be the first to weigh in.