oma-refactor — Claude Code skill: what it does, how to install
oma-refactor — a skill for Claude Code and Codex restructures existing code while preserving observable behavior. Executes targeted technical debt cleanup through small test-gated transformations.
MIT ★ 1,472 updated 2026-11-12
/plugin marketplace add first-fluke/oh-my-agent
/plugin install oma-security@oh-my-agentWhat oma-refactor does
The skill strictly separates behavior changes from refactoring: feature work or bug fixes must be handled separately, preventing accidental regressions during cleanup.
Its execution model relies on verified transitions. The agent first audits test coverage, determinism, and flakiness. If no safety net exists on legacy code, it writes characterization tests and commits them first before touching application logic. Each transformation runs individually; a repeated test failure triggers a Mikado revert of only that step.
Operation requires working test runners, code-intelligence symbol tools, and Git history for churn analysis. Plans and before/after metric deltas are saved to .agents/results/refactor/.
What it helps with
- break down complex files without altering observable behavior
- generate characterization tests for legacy brownfield code
- execute preparatory refactoring before adding new features
- reduce cyclomatic and cognitive complexity in hot spots
Example prompts for the agent
- Refactor
src/auth.tsinto smaller helper functions preserving behavior - Add characterization tests then decompose this legacy controller
- Prepare a step-by-step refactoring sequence for the order processing module
How to install oma-refactor
1Plugin marketplace
/plugin marketplace add first-fluke/oh-my-agent/plugin install oma-security@oh-my-agentThis installs the whole oma-security plugin with all its skills
2Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-refactor -a claude-code3Manually: put the files in place
~/.claude/skills/oma-refactor/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/oma-refactor/
1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-refactor -a codex2Manually: put the files in place
.agents/skills/oma-refactor/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-refactor -a cursor2Manually: put the files in place
.cursor/skills/oma-refactor/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install first-fluke/oh-my-agent .claude-plugin/domains/security/skills/oma-refactor2Manually: put the files in place
.github/skills/oma-refactor/SKILL.md1Skills CLI (Vercel)
npx skills add first-fluke/oh-my-agent --skill oma-refactor -a gemini-cli2Platform CLI
gemini skills install https://github.com/first-fluke/oh-my-agent --path .claude-plugin/domains/security/skills/oma-refactorManually: put the files in place
.devin/skills/oma-refactor/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about oma-refactor
How to install oma-refactor in Claude Code?
Run npx skills add first-fluke/oh-my-agent --skill oma-refactor -a claude-code in your terminal.
What does oma-refactor do?
It applies behavior-preserving structural improvements to your codebase, verifying every step with tests and recording clean separate commits.
How does oma-refactor handle code without tests?
It creates golden-master or characterization tests first, commits them separately, and only then starts modifying the source code.
Source text
SKILL.md · MIT
---
name: oma-refactor
description: Restructure existing code while preserving observable behavior. Use
for targeted technical debt or hotspot work with characterization tests.
---
# Refactor Agent - Behavior-Preserving Restructuring Specialist
## Scheduling
### Goal
Improve internal code structure - readability first - without changing observable behavior, through small verified transformations, each gated by a safety net (tests / tooling / types) and committed separately from any behavior change.
### Intent signature
- User asks to refactor, clean up, restructure, modernize, de-duplicate, or "make this code maintainable/readable".
- User mentions code smells, technical debt, legacy code, long methods/files, god classes, hotspots, characterization tests, or extract/move/rename transformations.
- User asks "where should we refactor first?" or wants a refactoring plan/priority for a codebase.
### When to use
- Executing a refactoring on specific files/modules (extract, move, rename, decompose, pattern/idiom alignment)
- Preparatory refactoring before a feature ("make the change easy, then make the easy change")
- Legacy (brownfield) rescue: seam discovery + characterization tests, then restructuring
- Refactoring target selection and prioritization (smells + SATD + hotspot = churn x complexity)
- Auditing whether code is safe to refactor now (coverage breadth x mutation strength x flakiness)
### When NOT to use
- Adding or completing behavior already requested by the user -> use the appropriate existing implementation skill; classify the scope under Entry first.
- Fixing a reported bug or failing behavior -> use `oma-debug` (refactoring must not change behavior)
- Security/performance/accessibility review or quality audit -> use `oma-qa`
- System design, module boundary decisions, ADRs, convention changes -> use `oma-architecture` (a convention/pattern change is an architecture decision, not a local refactoring)
- DB schema design or migration mechanics -> use `oma-db` (this skill only plans the expand-contract sequence)
- Commit splitting / staging mechanics -> use `oma-scm`
- Performance optimization as a goal -> out of scope by definition (tuning is a side effect, never the objective)
### Expected inputs
- `target`: file/module/path, smell report, SATD marker, or the feature request motivating preparatory refactoring
- `verification`: existing project test command(s); coverage/mutation tooling if already available
- `constraints`: coding guide / conventions, regulated-environment flags, merge-window concerns
- Optional: prior metric reports, hotspot data, ADRs touching the target area
### Expected outputs
- Refactored code as a sequence of atomic, refactor-only commits (no test changes mixed in)
- Safety-net additions when missing (characterization / golden-master tests) as separate commits
- Before/after report: metric delta (cyclomatic/cognitive complexity, size, coupling) + readability verdict
```yaml
outputs:
- name: report
description: refactoring plan or before/after report
artifact: ".agents/results/refactor/*.md"
required: false
```
Standalone runs write plan / before-after reports under `.agents/results/refactor/`; orchestrated runs (via the `refactor-engineer` agent) write `.agents/results/result-refactor[-{sessionId}].md` per the agent execution protocol.
### Dependencies
- `resources/definition.md` (invariant definition: 5 properties, boundaries, destination principle, naming roles, inline evidence)
- `resources/measurement.md` (4-layer measurement + git forensics commands)
- `resources/governance.md` (optional organization conventions and existing verification tools)
- Configured code-intelligence symbol/reference tools; existing project test runners for the target language
- Git history for churn/ownership/hotspot analysis
### Control-flow features
- Branches by safety-net state (greenfield vs brownfield), statefulness (code-only vs expand-contract), and verification outcome (pass vs Mikado revert)
- Reads code/history/metrics; writes code, tests (in separate commits), and reports
- Stops and routes to `oma-architecture` when the change requires a convention/boundary decision
## Structural Flow
### Entry
1. Classify the requested outcome against the actual target code, not the word "refactor" alone. Behavior-preserving restructuring stays here. Implementation or bug fixing already included in the request continues under the appropriate existing implementation skill or `oma-debug`, within existing authorization and without renewed approval. For mixed requests, separate behavior changes from restructuring and apply this skill only to the latter.
2. For the refactoring scope, establish its motivation and diagnose the safety net: coverage of changed lines, test determinism (flakiness), mutation strength if measurable.
3. Identify the destination form: the language idiom and codebase convention the result must match.
Missing or stub code does not itself authorize new behavior. A pure refactor with
no implementation to restructure reports the missing target rather than inventing
one. If a contract fact needed for correctness remains unknown after checking
available task and project evidence, ask for that specific fact under the shared
execution policy and clarification protocol. Distinguish missing
information from missing authorization: retain existing permission, pause only
dependent work, and continue independent authorized work.
### Scenes
1. **PREPARE**: Within the refactoring scope, classify greenfield (safety net exists) vs brownfield (build net first); check size gates and hotspot rank; confirm two-hats scope (no feature/bug work mixed in).
2. **ACQUIRE**: Read target code via symbol tools; collect metrics (complexity, size, coupling) and git signals (churn, ownership); read the coding guide for conventions.
3. **REASON**: Decompose the goal into a sequence of named atomic transformations; for stateful targets plan expand-contract; verify each step is independently verifiable and revertible.
4. **ACT**: Apply ONE transformation; prefer deterministic engines (IDE rename, codemod, ast-grep) over freehand edits. When one feature spans several files after a split, consider G-5's optional header map if it helps navigation.
5. **VERIFY**: Re-run existing tests unchanged. Pass -> commit (refactor-only) -> next transformation. Repeated failure -> Mikado: record the broken prerequisite, undo only this transformation's edits, recurse on the prerequisite first. Preserve pre-existing edits and other contributors' work.
6. **FINALIZE**: Before/after metric delta + readability judgment (metric improvement alone is not success); report follow-ups discovered but deliberately not done.
### Transitions
- If the safety net is missing or weak (low diff coverage, flaky, no assertions), write characterization / golden-master tests FIRST, committed separately, before touching production code.
- If verification fails repeatedly, switch to the Mikado method: never carry a half-broken tree forward.
- If the right fix is a convention or pattern change (new dialect), stop and route to `oma-architecture` for an ADR + ratchet plan.
- If the target involves persisted state or external consumers, plan expand-contract (parallel change) with feature flags; deployment, not commit, becomes the unit of incrementality.
- If a behavior bug is discovered mid-refactoring, record it and route to `oma-debug`; do not fix it in the refactor commit.
- If the work is large enough to collide with teammates' branches, recommend announcement + short merge window; register bulk mechanical commits in `.git-blame-ignore-revs`.
### Failure and recovery
| Failure | Recovery |
|---------|----------|
| Tests fail after a transformation | Mikado: record prerequisite, undo only this transformation's edits, preserve unrelated work, attack prerequisite first |
| No tests and code is untestable | Find a seam; apply only minimal mechanical changes to inject test access, then characterize |
| Tests are flaky | Fix or quarantine flaky tests before refactoring - an unreliable net is no net |
| Metric improves but readability worsens | Reject the transformation; readability is the success criterion, metrics are proxies |
| Scope keeps growing | Stop; report the boundary issue and split into a Mikado graph or route to architecture |
| Refactoring engine/codemod produces wrong output | Engines are not infallible - tests re-run is mandatory; fall back to manual atomic edits |
### Exit
- Success: behavior verified unchanged, structure measurably improved, readability confirmed, refactor-only commits, follow-ups reported.
- Partial success: safety net built but restructuring deferred; or prerequisites mapped (Mikado graph) with explicit blockers.
- Handoff: already-requested implementation or bug fixing continues under its owning skill's completion criteria; changing skills is not a task failure.
- Blocked or failed: report the exact missing contract, verification path, or required decision and the dependent work it prevents, plus independent work completed. A pure refactor with no target reports that limit without inventing behavior.
## Logical Operations
### Actions
| Action | SSL primitive | Evidence |
|--------|---------------|----------|
| Diagnose safety net | `VALIDATE` | Coverage/flakiness/mutation state of target scope |
| Collect signals | `READ` | Metrics, git churn/ownership, smells, SATD |
| Rank targets | `COMPARE` | Hotspot = complexity x churn |
| Plan atomic sequence | `INFER` | Named transformations, Mikado graph |
| Write characterization tests | `WRITE` | Golden-master/snapshot tests (separate commit) |
| Apply transformation | `WRITE` / `CALL_TOOL` | One atomic refactor, engine-first |
| Verify preservation | `VALIDATE` | Existing tests re-run unchanged |
| Commit separately | `UPDATE_STATE` | `refactor:`-typed commits only |
| Report delta | `NOTIFY` | Metric + readability before/after |
### Tools and instruments
- Configured code intelligence for symbol/reference/pattern impact analysis; an available semantic rename engine for renames. Native inspection remains valid, but do not replace a semantic rename with blind text replacement
- Deterministic transformers: IDE refactoring actions, codemods (jscodeshift / OpenRewrite / ast-grep / comby)
- Metrics: use already installed complexity tools such as lizard / radon and existing project lint limits; installing or downloading additional tools requires authorization
- Test stack: preserve the project runner and configuration; use mutation tooling only when available and appropriate (see `resources/governance.md`)
- Git forensics one-liners (see `resources/measurement.md`)
### Canonical workflow path
1. Apply Entry's scope routing first. For the remaining refactoring scope, run coverage and check test determinism; classify green/brownfield.
2. If brownfield: find a seam, write characterization (golden-master) tests for CURRENT behavior, commit.
3. Select targets by hotspot rank (complexity x churn), not by smell aesthetics alone.
4. Plan a sequence of named atomic transformations toward the language-idiomatic, convention-conforming form.
5. Loop per transformation: apply (engine-first) -> re-run tests UNCHANGED -> commit `refactor:` only.
On repeated failure: record prerequisite, undo only this transformation's edits,
preserve pre-existing and concurrent work, recurse (Mikado).
6. Finish: metric delta + readability verdict; list discovered-but-deferred work; never mix in behavior changes.
### Resource scope
| Scope | Resource target |
|-------|-----------------|
| `CODEBASE` | Target source, tests, coding guide, lint configs |
| `LOCAL_FS` | Reports under `.agents/results/refactor/`, `.git-blame-ignore-revs` |
| `PROCESS` | Test runners, coverage/mutation tools, codemod engines, git log analysis |
| `MEMORY` | Mikado prerequisite graph, deferred follow-ups, metric baselines |
### Preconditions
- A verification path exists or can be built (tests/types/tooling); otherwise the first deliverable is the safety net, not restructuring.
- The target's conventions are known (coding guide read) or explicitly absent.
### Effects and side effects
- Mutates production code (structure only) and adds tests in separate commits.
- Runs test/coverage/mutation commands; reads git history.
- May write reports under `.agents/results/refactor/` and entries to `.git-blame-ignore-revs`.
- Never alters observable behavior, public contracts, or persisted data without an expand-contract plan.
### Guardrails
1. **Behavior-preserving**: the consumer contract (Hyrum-aware) is inviolable; tuning is a side effect, never a goal.
2. **Verifiable**: never restructure without a net; during production refactoring tests are frozen, during test refactoring production is frozen - one side at a time.
3. **Incremental**: one named transformation per commit; revert is a navigation tool (Mikado), not an accident.
4. **Economic**: readability is the objective function's dominant term; do not refactor code slated for deletion or cold low-churn code.
5. **Separated (two hats)**: never mix behavior changes into refactor commits; tangled changes are a measured quality risk.
6. Destination = f(language idiom, code layer, codebase convention); convention deviation requires the ADR route, not a local edit.
7. Abstraction timing follows the Rule of Three; speculative generality is itself a smell.
8. All metrics are proxies (Goodhart): a 499-line mechanical split, assertion-free coverage, or pattern-count gains are failures, not wins.
## References
- Local code tools: `../_shared/core/code-intelligence.md` (code search/navigation)
- Authorization and contract ambiguity: `../_shared/core/execution-policy.md` and `../_shared/core/clarification-protocol.md` (when clarification or authorization is needed)
- Invariant definition (5 properties, boundaries, destination, naming roles, contexts, D&C, inline evidence): `resources/definition.md`
- Measurement: 4 layers + git forensics commands: `resources/measurement.md`
- Optional organization conventions and existing verification tools: `resources/governance.md`
- Context loading: `../_shared/core/context-loading.md`
- Quality principles: `../_shared/core/quality-principles.md`
- Adjacent skills: `oma-debug` (bugs), `oma-qa` (audits), `oma-architecture` (boundaries/ADR), `oma-db` (schema), `oma-scm` (commits)
© First Fluke · MIT · first-fluke/oh-my-agent