analyze

name	analyze
description	Visible Harness analyze entry. Use after experiment runs to turn results into interpretation, decisions, and next-run or writing guidance.

Analyze

Use $analyze as the human-facing alias for result analysis. This is not a separate Skill Contract.

Read and follow:

• ../../../.agents/skills/evaluate/SKILL.md
• ../../../.agents/references/workflow-guide.md
• ../../../AGENTS.md
• ../../../CLAUDE.md

Analyze current run artifacts and iteration logs, then produce decision-ready insights that guide the next $run, a $grill pivot, or $write.

After updating completed run evidence, refresh docs/30_evidence/Experiment_Evidence_Index.{json,md} using tooling/evidence/build_experiment_evidence_index.py so $write can consume experiment results without treating iteration_log.json as the sole source of truth. If not run, report NOT_RUN with the reason.