
from spec-gen105
Apply a confirmed refactoring plan safely: reads a refactor plan, applies changes as small, test-gated mini-developments, and stops on failures to avoid lost co
Executes a refactoring plan produced by spec-gen-plan-refactor. It treats each planned change as an independent mini-development cycle: read the plan entry, make a targeted edit (≤50 lines), diff to verify no lost code, run the specified test gate, and mark the change done only when tests pass. Includes circuit-breaker rules and safety constraints to avoid regressions.
Use when you have a confirmed refactor plan in .spec-gen/refactor-plan.md and need an automated, disciplined executor to apply changes reliably across a codebase with strict testing and rollback rules. Not for ad-hoc edits — only for plan-driven refactors.
Fits agents that can run shell/git commands and tests (CI-enabled agents, developer assistants, spec-gen MCP tooling).
Skill instructs agents to apply a pre-confirmed refactoring plan from .openlore/refactor-plan.md using a disciplined mini-development loop (read→edit→diff→test). No bundled scripts — purely instruction-driven, relying on openlore MCP tools and git. Well-structured with circuit-breaker logic, coverage thresholds, and diff verification to prevent lost code. Requires openlore MCP server as prerequisite, limiting standalone use.
Original GitHub URL (clay-good/spec-gen) redirects to clay-good/OpenLore — the skill was renamed from spec-gen-execute-refactor to openlore-execute-refactor. SKILL.md content fetched from new repo location. No scripts to audit. Strong safety-first design with multiple guardrails against broken state accumulation.