auto-improvement

name	auto-improvement
description	Focused measurement-aware refinement loop for Robot SF prompts, docs, and small code changes; use when a task benefits from trying a few simple improvements.
category	research-iteration
kind	atomic
phase	analysis
requires_write	true
requires_slurm	false
requires_benchmark_artifacts	false
delegates_to	[]
output_schema	skill_run_summary.v1

Auto Improvement

When to use

Use this skill for small, iterative refinements that should improve clarity, robustness, or validation coverage without launching full experimentation campaigns.

If the task needs repeated measurable experiments, use autoresearch instead.

Read First

• AGENTS.md
• docs/dev_guide.md
• .specify/memory/constitution.md
• docs/ai/awesome_copilot_adaptation.md

Workflow

1. Identify the smallest measurable improvement target.
2. Pick one validation signal and one narrow change at a time.
3. Limit scope to the named files.
4. Apply one tweak; validate; keep or revert quickly.
5. Stop when gain is too small for complexity or the user stops.

Proof and Guardrails

• Keep changes small and reversible.
• Keep benchmark/provenance wording conservative.
• Prefer scripts/dev/ wrappers for validation.
• For tasks that are plain cleanup/formatting, use clean-up.
• If validation shows no net gain, revert.

Good Fits

• tightening prompt wording,
• improving docs discoverability,
• reducing duplication in a small helper,
• adding a narrow regression test,
• clarifying a workflow note.

Output

Summarize:

• what changed,
• what improved,
• what remains open.

Guardrails

• Stay within the skill scope declared in .agents/skills/skills.yaml.
• Prefer repository scripts and canonical docs before ad-hoc commands.
• Record blockers and validation gaps instead of overstating completion.