Category

Harnesses

5 pages in this section.

Agent Eval Gate Agent Eval Gate Overview A release gate for coding or tool using agents that checks task success, policy adherence, regression risk, and human review handoff before broader rollout. Conference Context The Evals track tre Coding Agent Code Review Loop Coding Agent Code Review Loop Overview A workflow for letting agents produce code while preserving reviewer comprehension, test evidence, rollback paths, and issue specific context. Conference Context Software factory an Context Graph Ingest Context Graph Ingest Overview A harness for turning documents, talks, repositories, profiles, and transcripts into a source labeled graph that agents can retrieve without losing provenance. Conference Context Context gra Sandboxed Agent Execution Sandboxed Agent Execution Overview A runtime boundary for agents that can run code, touch files, browse, call tools, or influence production systems. Conference Context Sandbox talks argue that agent reliability depends Transcript Slide Synthesis Loop Transcript Slide Synthesis Loop Overview A repeatable way to combine schedule metadata, transcripts, slide OCR, reconstructed slides, and topic pages without promoting weak evidence into confident claims. Conference Cont