Summary

• Adds make eval-{task-new,record,review,compare,list} targets for capturing flow runs as versioned, reviewable artifacts.
• A run is pinned to kernel SHA + active-cell SHA, copies the thread's handoff docs, captures the diff against the base branch, and writes a JSON manifest.
• The review template is markdown with rich qualitative sections (overall impression, per-stage notes, doc readability for both human + machine readers, code quality, and a "patterns flow should learn" section). Numeric scores are optional.
• Reviews live in evals/runs/<task>/<run-id>/reviews/<reviewer>.md and are committed alongside the run, so /flow:reflect (and humans iterating on flow) can read them next time.
• eval-compare prints version pins and metric deltas side by side and points at shared reviewers' files for prose-level diffing.

Layout

evals/
  tasks/<task-id>/             # task definitions
  runs/<task-id>/<run-id>/     # captured runs (manifest + thread + diff + reviews)
  templates/                   # review template + task README scaffold

run-id = <utc-iso>-<kernel-sha7> — sortable, scannable.

Test plan

• make eval-task-new TASK=<id> scaffolds tasks//{task.json, prompt.md, README.md}
• make eval-record TASK=<id> PROJECT=<path> discovers the most recent thread, copies it, generates the diff, writes a manifest pinned to current kernel + cell SHAs
• make eval-review TASK=<id> RUN=<run-id> opens $EDITOR on a pre-filled review template with an artifact index appended
• make eval-compare TASK=<id> A=<a> B=<b> prints version + metric deltas and lists reviews
• make eval-list summarises tasks and runs

🤖 Generated with Claude Code