Refactor safely

Skill contextosai/skills/skills/refactor-safely

A collection of Agent Skills — self-contained folders that teach AI agents to perform specialized tasks. Includes a production-readiness harness-audit skill, a plugin marketplace, a SKILL.md spec, and CI validation.

Install

npx -y skills add contextosai/skills --skill refactor-safely

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Restructure existing code while preserving externally observable behavior through explicit invariants, characterization evidence, small reversible steps, and continuous verification. Use when the user asks to refactor, simplify, extract, reorganize, modularize, deduplicate, split a component, improve architecture, or pay down technical debt without intentionally changing product behavior. Distinguish refactoring from feature work and stop scope drift from hiding inside structural edits.

SKILL.md

2.6 KB, as published. Nobody here has run it

Refactor Safely

Change structure while keeping the behavior contract stable.

Protocol

Define the refactor boundary and list observable behavior that must not change: API, side effects, ordering, errors, timing-sensitive semantics, persistence, telemetry, and extension points.
Trace callers and consumers across the boundary. Include reflection, registration, configuration, serialization, and framework conventions that symbol search may miss.
Assess existing proof. If behavior is weakly specified, add the smallest characterization tests around high-risk invariants before restructuring. Do not encode accidental behavior without evaluating whether callers rely on it.
Choose a seam that permits incremental movement. Prefer mechanical moves, then semantic cleanup; do not combine both when the diff would obscure behavior changes.
Execute one reversible transformation at a time. Keep old and new paths from diverging during transition; use adapters or delegation briefly when needed.
Verify after each meaningful step with focused checks, then run boundary and broader tests proportional to risk.
Compare before/after public surface, generated artifacts, state transitions, error behavior, and performance where relevant.
Remove transitional code only after all callers migrate and proof covers the new path. Search for stale imports, registrations, flags, and documentation.

Guardrails

Do not sneak bug fixes or feature changes into a behavior-preserving claim. Split and label intentional behavior changes.
Do not equate compilation with behavioral equivalence.
Do not introduce an abstraction before identifying at least one stable invariant it owns.
Avoid wrappers that merely relocate complexity or duplicate sources of truth.
Preserve user edits and repository conventions; prefer the smallest reviewable sequence over a sweeping rewrite.

Output

For planning-only requests, use references/refactor-ledger.md. For implementation requests, maintain the ledger while editing and report preserved invariants, intentional deviations, checks run, and residual risks.