complexity-mitigator
Existing-code comprehension and local winnowing preflight. Use for simplify/refactor/clean up/untangle, nested branches, boolean soup, opaque names, mixed responsibilities, cross-file state, or review stalls. Factor the local whole, separate essential/incidental/specification-risk factors, winnow dominated or duplicated factors, and emit the smallest clarity cut. Not for broad architectural layer removal, kernel quotienting, invariant remediation, or greenfield planning.
How do I install this agent skill?
npx skills add https://github.com/tkersey/dotfiles --skill complexity-mitigatorIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The complexity-mitigator skill is a code analysis tool designed to help developers comprehend and simplify complex existing code. It is strictly read-only and includes explicit guardrails that prohibit any file modifications or commits. The skill does not use any external dependencies, network operations, or privileged commands. No malicious patterns or obfuscation techniques were detected during analysis.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
- Runlayerwarn
1/1 file flagged
What does this agent skill do?
Complexity Mitigator
Intent
Reduce understanding cost in existing code while preserving essential domain meaning. This skill is a read-only routing rail and analysis preflight, not a delivery owner.
FACTORING -> WINNOWING
- Factor the code into responsibilities, states, effects, decisions, and boundaries.
- Retain essential domain factors.
- Expose specification-risk factors before refactoring.
- Remove, collapse, rename, localize, or delegate incidental factors.
- Route repeated semantic distinctions or causal review pressure to
$review-foldand$actuating.
Activation
Use for hard-to-follow existing code, deep nesting, boolean soup, mixed parse/validate/decide/effect behavior, cross-file hops, hidden state, ordering dependencies, opaque names, and review stalls caused by comprehension.
Handoff instead:
- broad framework or layer tax ->
$reduce; - unclassified review evidence ->
$review-fold; - classified owner-boundary pressure ->
$actuating; - illegal states or invariant ownership ->
$invariant-ace; - missing essential structural shape ->
$universalist; - implementation ->
$actuatingor the owning workflow.
Workflow
- Choose the slice: entry point, inputs, outputs, state, effects, and boundary.
- Factor it into responsibilities, decisions, state, effects, external obligations, and proof surfaces.
- Classify each factor as
essential,incidental,mixed, orspec-risk. - Identify duplicated, dominated, subsumed, vestigial, pass-through, or misplaced factors.
- Preserve essential factors and unresolved specification risk.
- Winnow in this order: delegate, flatten, rename, localize, collapse, then extract when a coherent design decision or lifecycle can be hidden behind a simpler sufficient contract. Single-use encapsulation needs no second caller; reusable generalization across cases needs evidenced commonality.
- State the recomposition rule and smallest proof signal.
- Return a handoff; do not implement.
Actuating composition
Within Actuating, the checked-in complexity review lens owns the lane's scope and return shape. Its focus is unnecessary correctness machinery and duplicate semantic ownership, not the full standalone comprehension preflight. Return the independently maintained truth, accepted obligation, and evidence; distinguish competing owners from derived guards and independent oracles. Keep grounded nonblocking opportunities separate from established violations. Do not select a repair, launch companion reviews, or emit a second preflight, evidence table, or proposed-repair agenda. Review Fold adjudicates and Actuating owns the construction decision. This section takes precedence over standalone workflow, routing, and output only within Actuating. Standalone behavior is unchanged.
Output
Complexity preflight:
- whole:
- dominant cost:
- factorization:
- retain:
- remove/collapse/delegate:
- unresolved specification risk:
- recomposition:
- proof:
- handoff:
Guardrails
- Do not edit files or commit.
- Do not select architecture, review credit, mutation, or closure.
- Do not confuse fewer lines with lower understanding cost.
- Do not delete essential policy or unresolved external obligations.
- Do not extract for hypothetical reuse or before a coherent shape is visible. Name what the caller can stop knowing or doing; a pass-through helper or a configuration facade that retains the same obligations is not a clarity cut.
- If behavior is unclear, request an executable learning surface such as an example matrix, contract test, fixture set, or state table.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/tkersey/dotfiles/complexity-mitigator">View complexity-mitigator on skillZs</a>