Article library
268 articles · Page 3
Keep browsing
Use tags to narrow the library once you know the direction you want.
A practical observability model for reconstructing agent tasks across model calls, tool activity, state changes, costs, and failures while minimizing sensitive prompt and user data.
Read articlearrow_forward
A failure-recovery loop for agentic coding interviews: stop compounding damage, preserve the failure signal, choose the correct recovery boundary, and restart with a smaller falsifiable change.
Read articlearrow_forward
A practical permission-gate framework for coding-agent interviews: classify the action, inspect its blast radius, narrow risky commands, and explain the approval with a rollback and verification plan.
Read articlearrow_forward
A deep system-design answer for governing agent execution by separating task intent, policy, credentials, workers, approvals, state, compensation, and audit so consequential actions stay bounded and inspectable.
Read articlearrow_forward
A defensible provenance method for AI-assisted take-homes: follow the employer's policy, record material contributions, show your modifications and verification, disclose concisely, and avoid retaining confidential assignment data.
Read articlearrow_forward
A careful reading of Greenhouse's U.S. 70% and 8% findings: they show a perception gap between hiring-manager confidence and job-seeker fairness, not proof of AI accuracy, discrimination, or global opinion.
Read articlearrow_forward
A practical, U.S.-context guide to identifying disability-related barriers in AI-enabled interviews, requesting an effective adjustment, confirming the agreed process, and preserving only the information needed.
Read articlearrow_forward
A practical evidence chain for permitted AI use: preserve the material request, connect it to the reviewed code change, record relevant verification, and state the human choice without exposing confidential content.
Read articlearrow_forward
A candidate-facing checklist for reading AI interview notices before pressing Start: identify each data layer, its purpose and recipients, retention and deletion terms, jurisdiction-specific rules, and possible alternatives.
Read articlearrow_forwardPage 3 of 30
