What it is
A consolidated field guide to seven harness-shaped efforts in this workspace — including the Sam Queue delegation and learning control plane — each mapped to its directory, its mechanism, and its status.
Harness Atlas
One map of every test, benchmark, and eval harness in the workspace — what each is, which directory it lives in, and where they overlap.
Read the introduction Jump to the overlap map
Tease:Seven harness-shaped efforts, one honest map.Lede:Harness Atlas compares execution, review, and operational learning systems without pretending they are one product.Why it matters:Shared evidence patterns and real lifecycle status make consolidation decisions reviewable.Go deeper:Start with the introduction or jump directly to the overlap map.
Why it exists
These grew independently across two years and several domains. They share a spine — run a thing → capture evidence → score it deterministically — that’s easy to miss when each lives in its own folder. This is the map that makes the overlap visible.
The seven harnesses
Section titled “The seven harnesses” agentlab Overnight local-LLM coding-agent benchmark, sandboxed in Docker. Working.
sam-evals The planned public extraction of agentlab's engine. Planned.
doesitarm-harness Mac-app compatibility tester — drives apps, records evidence, classifies arch. Prototype.
roof-frames Radar-frame benchmark — renders frames, pixel-scores them vs TV/NOAA. Prototype.
harness-watch Weekly radar on the external agent-harness ecosystem. Active.
Sam Queue Bounded delegation, trace analysis, memory, and eval-driven improvement. Prototype.
work-review automation Read-only internal feedback loop for automation opportunities. Planned.
Start here
Section titled “Start here” Introduction What this atlas is and the question it answers.
The shared spine The run → capture → score shape and feedback loops all seven share.
Comparison matrix All seven, side by side, on one table.
Overlap & shared primitives Where they actually overlap — and what could merge.