simplify-code

name	simplify-code
description	Use when code feels over-engineered, rotted, or duplicated — cut unused abstraction, inline single-use helpers, centralize patterns repeated in 3+ places, prefer structural impossibility over defensive clutter. Behavior-preserving. Phase = Simplify.
when_to_use	when reviewing existing code that looks bloated, when a refactor request lands, when the same pattern shows up in 3+ places, or when a single-use abstraction is adding cost without payoff
tier	1
phase	simplify

Simplify Code

Simplify-phase skill — cross-phase, usable during Build (refactor intent) or as standalone cleanup. Cut complexity that does not earn its keep. Behavior-preserving — every change is provable by the existing tests.

Iron Rule

1. NEVER simplify without a test suite that proves behavior before and after. 2. NEVER remove an abstraction the codebase actually depends on — verify call sites first. 3. NEVER add a new abstraction for "hypothetical future use". One concrete user is not enough. 4. Same pattern in 3+ places → centralize. On the high-risk list — auth, billing, credits, URL validation, redirects, SSRF, cookies, logging, retries, external API — TWO occurrences already force it. 5. Apply the deletion test before any cut. Imagine deleting the module: if complexity vanishes, it was a pass-through — delete safely. If complexity reappears scattered across N callers, the abstraction was earning its keep — keep it.

When to use

• Code reviewer flagged over-engineering or duplication
• A file is > 500 lines and looks like it grew by accretion
• An abstraction has exactly one caller
• A defensive null-check / try-catch covers an "impossible" case
• Same logic copy-pasted in 3+ files
• User says "this is getting messy" or "refactor X"

Skip when:

• Tests don't exist for the touched code — write them first via implement-plan or debug-issue
• The "complexity" is load-bearing (security boundary, data invariant)
• It is mid-feature; finish the feature first

Boundary

Owns:

• Behavior-preserving complexity cuts: inline single-use helpers, delete unused config, centralize repeated patterns, structural simplification.

Does not own:

• Feature changes.
• Bug fixes with unknown root cause.
• Refactors without a test baseline.
• Product / API behavior changes.

Return / hand off:

• Behavior must change → write-spec or write-plan.
• Tests missing for a risky area → implement-plan to add baseline tests.
• Bug found while simplifying → debug-issue.
• Simplification complete → check-work.

Inputs to gather

• The code region or file(s) flagged as complex
• The existing tests for that region (must be green before starting)
• Call sites for any abstraction you plan to inline or remove
• The user's intent (cleanup only, or cleanup + behavior change)

Workflow

1. Confirm tests are green

Run the touched module's test suite. If red, fix or write tests first. You cannot prove behavior-preserving without a baseline.

2. Scan for these patterns

Before removing anything, run Chesterton's Fence + the deletion test (Iron Rule 5) together. Chesterton: git blame the origin commit — code with no callers may still encode a reason. Verify the why, not just the call sites. Both tests pass = safe to delete; either fails = stop.

Pattern	Action
Interface / type with one implementation	Inline the impl, delete the interface
Config flag with one value used in code	Delete the flag
Helper / wrapper with one caller	Inline at the call site
Retry / timeout config without observed failure	Delete; add back when a real failure appears
Defensive null check on a value that cannot be null structurally	Tighten the type, delete the check
Same 5-line pattern in 3+ files	Extract to one source of truth
Backwards-compat shim for code nobody calls	Delete the shim
Comment that restates what the code does	Delete the comment
Wrapper that only forwards calls (delete it → complexity vanishes)	Inline; a pure pass-through earns nothing

3. Prefer structural over runtime

A runtime if (x === null) throw becomes a non-nullable type. A "must be set" config becomes a required constructor argument. Make the bad state un-representable when the type system allows.

4. Centralize at 3 occurrences

Two is a coincidence. Three is a pattern — centralize. On Iron Rule 4's high-risk list (auth / billing / credits / URL validation / redirects / SSRF / cookies / logging / retries / external API), two occurrences already force it.

Inverse rule for keeping abstractions. One adapter behind an interface = hypothetical seam, candidate for inline. Two real adapters = real seam, keep the interface. The rule mirrors centralization: counts decide structure. An interface with one implementation is the same shape as a five-line pattern in one file — it does not yet earn the abstraction.

4b. Refactor before fix

When a planned change is hard because the surrounding shape is wrong, first cut the shape until the change is easy, then make the easy change. Two commits, not one. Cut commits are behavior-preserving (this skill); the change commit is the feature (implement-plan). Mixing them in one commit makes the diff unreviewable and hides which line caused which regression.

Skip this rule if the change is genuinely small and the shape is fine; do not invent friction to justify a refactor.

5. One change at a time

One cut per commit. Run the test suite between cuts. A failing test mid-simplification tells you the previous cut went too far — revert that one, not all of them.

6. Stop when behavior is at risk

If a cut requires changing a test assertion to keep it green, you are no longer behavior-preserving. Stop, ask the user, or move the cut to a separate implement-plan task with a real spec.

If a matching Rolepod agent is available

Delegate to the closest specialist:

• universal-reviewer for DRY / smell / structure cleanup
• system-architect when the cut touches module boundaries or APIs
• security-engineer when the cut touches auth / secret / token / crypto code paths

Brief: the file region, the existing tests, the user intent (cleanup vs cleanup + behavior).

If no matching agent is available

Execute as Lead with this minimum viable checklist:

1. Run the module's test suite — must be green
2. List the over-engineering patterns present in the file
3. Pick the smallest single cut
4. Apply it
5. Run the test suite — must stay green
6. Commit (or stage); repeat with the next smallest cut
7. Stop when further cuts would change behavior or break tests
8. Centralize anything that appears in 3+ files into one source of truth

Output

The simplification report is the canonical artifact: templates/simplification-report.md. It carries the green baseline, each cut, anything centralized, the post-cut tests, and the behavior-preserved verdict. Do not restate the report shape here; the template is the single source.

Examples

Non-blocking — read only when unsure whether a cut is behavior-preserving:

• examples/simplify-examples.md — a single-use-helper inline and a defensive-check cut, each a good/bad pair with a "why good wins" table. Read the whole file; the contrast is the lesson.

Hard stops

• Tests were not green at the start → write tests first, do not simplify on a red baseline
• A cut required a test assertion change → that is a behavior change, route to implement-plan
• An "unused" abstraction has callers you missed → restore it and verify before another attempt
• About to invent a new abstraction for one caller → reject; that is complexity, not simplification
• About to delete a module without running the deletion test → stop; Iron Rule 5
• Refactor-before-fix commits mixed with the feature change in one commit → stop, split; cut commits stay behavior-preserving

Full Rolepod enhancement

Full Rolepod improves this phase by adding the simplicity gate S1-S5 in pre-commit hooks, universal-reviewer adversarial pressure on DRY violations, and the centralization rule enforced for auth / billing / credit / URL validation paths.

Next phase

• If the cleanup is part of a larger plan, return to implement-plan for the next planned task.
• If the cleanup uncovered a real bug, route to debug-issue.
• If the cleanup is complete, route to check-work for the verification block, then finish-work.