Paths: File paths (shared/, references/, ../ln-*) are relative to skills repo root. If not found at CWD, locate this SKILL.md directory and go up one level for repo root.

ln-813-runtime-optimizer

Type: L3 Worker Category: 8XX Optimization Parent: ln-810-performance-optimization-coordinator

Fixes runtime performance issues found by ln-653-runtime-performance-auditor. Each fix verified via tests + lint with keep/discard pattern.

Overview

Aspect	Details
Input	Audit findings from `docs/project/persistence_audit.md` (ln-653 section) OR target file
Output	Optimized code, verification report
Companion	ln-653-runtime-performance-auditor (finds issues) → ln-813 (fixes them)

Workflow

Phases: Pre-flight → Load Findings → Prioritize → Fix Loop → Report

Phase 0: Pre-flight Checks

Check	Required	Action if Missing
Audit findings OR target file	Yes	Block optimization
Test infrastructure	Yes	Block (need tests for verification)
Linter available	No	Skip lint step in verification
Git clean state	Yes	Block (need clean baseline for revert)

MANDATORY READ: Load shared/references/ci_tool_detection.md — use Test Frameworks + Linters sections.

Worktree & Branch Isolation

MANDATORY READ: Load shared/references/git_worktree_fallback.md — use ln-813 row.

Phase 1: Load Findings

From Audit Report

Read docs/project/persistence_audit.md, extract ln-653 findings:

Finding Type	Optimization
Blocking IO in async	`asyncio.to_thread()`, `aiofiles`, async HTTP client
String concat in loop	`StringBuilder` (.NET), `"".join()` (Python), template literals (JS)
Unnecessary allocations	Pre-allocate collections, object pooling, `Span<T>`
Sync sleep in async	`asyncio.sleep()`, `Task.Delay()`, `setTimeout`
Redundant serialization	Cache serialized form, avoid round-trip JSON in memory

From Target File

If no audit report: scan target file for runtime patterns matching the table above.

Phase 2: Prioritize Fixes

Priority	Criteria
1 (highest)	Blocking IO in async context (blocks event loop)
2	Sync sleep in async (starves thread pool)
3	Allocations in hot loop (GC pressure)
4	String concatenation patterns
5	Other micro-optimizations

Phase 3: Fix Loop (Keep/Discard)

Per-Fix Cycle

FOR each finding (F1..FN):
  1. APPLY: Edit code (surgical change)
  2. VERIFY: Run tests
     IF tests FAIL → DISCARD (revert) → next finding
  3. VERIFY: Run lint (if available)
     IF lint FAIL → DISCARD (revert) → next finding
  4. BOTH PASS → KEEP
  5. LOG: Record fix for report

Keep/Discard Decision

Condition	Decision
No tests cover affected file/function	SKIP finding — log as "uncovered, skipped"
Tests + lint pass	KEEP
Tests fail	DISCARD + log reason
Lint fail (new warnings)	DISCARD + log reason
Lint unavailable	KEEP (tests sufficient)

Phase 4: Report Results

Report Schema

Field	Description
source	Audit report path or target file
findings_total	Total findings from audit
fixes_applied	Successfully kept fixes
fixes_discarded	Failed fixes with reasons
fix_details[]	Per-fix: finding type, file, before/after description

Configuration

Options:
  # Source
  audit_report: "docs/project/persistence_audit.md"
  target_file: ""

  # Verification
  run_tests: true
  run_lint: true

  # Scope
  fix_types:
    - blocking_io
    - sync_sleep
    - unnecessary_alloc
    - string_concat
    - redundant_serialization

Error Handling

Error	Cause	Solution
No audit findings	ln-653 not run or no issues	Report "no findings to optimize"
Async rewrite complex	Deep call chain affected	Log as manual step, skip
Framework-specific pattern	Unknown async framework	Query Context7/Ref for docs

References

../ln-653-runtime-performance-auditor/SKILL.md (companion: finds issues)
shared/references/ci_tool_detection.md (test + lint detection)

Definition of Done

Findings loaded from audit report or target file scan
Fixes prioritized (blocking IO first, then sync sleep, allocations, string concat)
Each fix applied with keep/discard: tests + lint pass → keep, either fails → discard
Report returned with findings total, fixes applied, fixes discarded

Version: 1.0.0 Last Updated: 2026-03-08