Index 04
What I test
Experiments.
Working notes from ideas under test: why I started, what changed, what failed, and whether the result is worth continuing.
Keeping coding-agent work alive after the terminal closes
Testing whether a persistent runtime can let TUI, Web, and CLI clients share task state and recover approvals, tool output, diffs, and verification after an interruption.
Routing Codex across incompatible model protocols
Testing whether a local gateway can preserve the Codex workflow while handling protocol conversion, model routing, failover, and request tracing.
Reducing noise in AI code review
Testing whether parallel review, deduplication, independent judging, and confidence filtering can keep high-risk issues in front of a merge while reducing low-value feedback.