メニュー
MySQL schema design, indexing, and query tuning. Use when modeling tables, choosing types/keys, writing or optimizing SQL, reading EXPLAIN output, or resolving slow queries, deadlocks, and transaction issues.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Apache2 (httpd) configuration and ops for PHP services — vhosts, MPM tuning, PHP-FPM via mod_proxy_fcgi, mod_rewrite, TLS/HTTP2, compression/caching, hardening, and diagnosing 5xx/perf issues. Use for web-server config, deployment topology, and Apache performance/security.
Test automation strategy and engineering across the pyramid — unit, integration, API/contract, and end-to-end. Use for designing a test strategy, building API/E2E suites, CI test pipelines, fixtures/factories, flake reduction, and coverage of a REST microservice. Complements phpunit-testing for cross-cutting and higher-level testing.
Modern PHP 8.4/8.5 authoring and review. Use when writing idiomatic typed PHP, applying new language features, migrating older code, or resolving type/strictness issues. State the minimum PHP version any recent feature requires.
PHP performance — profiling, removing hotspots, cutting latency/memory, OPcache/JIT/preloading, eliminating N+1, efficient structures, generators/streaming, connection reuse, benchmarking. Use when something is slow or memory-heavy, or to validate an optimization. Measure first.
Redis from PHP — caching, sessions, rate limiting, locks, queues/streams, pub/sub, sorted-set patterns, TTL/eviction. Use when adding Redis to a feature, designing key schemes, or fixing cache correctness/stampede issues.
Defensive application security for PHP/web services — secure coding review, input validation, authn/authz, session/token security, injection/XSS/CSRF/SSRF/IDOR/deserialization defenses, secrets, and crypto. Use for security reviews and security-sensitive design. Defensive use only.
| name | mysql-expert |
| description | MySQL schema design, indexing, and query tuning. Use when modeling tables, choosing types/keys, writing or optimizing SQL, reading EXPLAIN output, or resolving slow queries, deadlocks, and transaction issues. |
InnoDB always. Narrowest correct type (DECIMAL for money, never FLOAT; utf8mb4; deliberate DATETIME vs TIMESTAMP). Explicit compact PKs (monotonic surrogate for write-heavy tables). FKs unless a deliberate reason not to. Normalize first; denormalize only with a measured reason.
Design from real WHERE/JOIN/ORDER BY patterns. Know leftmost-prefix, composite column order, covering indexes, selectivity. Functions on a column disable index use. Avoid redundant/over-indexing (write cost).
Reason from EXPLAIN / EXPLAIN ANALYZE. Kill full scans on hot paths, filesort/temp tables, and N+1. No SELECT * in app code. Keyset (seek) pagination over large OFFSET.
Know isolation levels (default REPEATABLE READ), gap/next-key locks, deadlock causes (order operations consistently), keep transactions short. Write reversible, online-safe DDL; flag locking operations on large tables (gh-ost/pt-osc).
Request SHOW CREATE TABLE + the query + EXPLAIN before prescribing. Give the concrete index/DDL/rewrite, explain why via optimizer behavior, and state the trade-off.
