dd3ok

Audits stale, unsafe, contradictory, or bloated agent-facing context and produces cleanup plans. Use for prompt debt, obsolete agent instructions, memory cleanup, source-of-truth drift, AGENTS.md-style docs, and skill descriptions, 프롬프트 부채, and 컨텍스트 정리. Do not use for ordinary README edits, general code refactors, linting, summarization, or creating a memory bank.

Reviews archived Hermes skill context.

Reviews Hermes migration context for stale instructions.

Review OpenClaw skill changes in this fixture.

Review default skill changes in this fixture.

Test Codex home skill fixture.

Test Codex plugin skill fixture.

Test Hermes top-level skill fixture.

Use this skill when the user asks to collect, validate, or interpret a redacted static inventory of Spring build files, declared versions, configuration keys, source/test signals, modules, or deployment artifacts, or to implement or validate a spring-evidence/1 consumer. A review or upgrade skill may hand off here when it identifies missing repository facts; missing facts alone do not make this the initial owner. Do not use it for dependency resolution, an upgrade plan, or an ordinary Spring explanation. Bundled scripts require Python 3.12+; build execution requires separate authorization and configuration values remain redacted.

Use this skill when the user asks to create a Spring Boot application or implement, refactor, remediate, or add code, configuration, tests, migrations, integrations, dependencies, or build declarations in an existing Spring project, with working repository changes as the primary output. Use it for greenfield scaffolding, including exact supported-version verification needed by that implementation, and bounded existing-project changes. Do not use it for spring-evidence/1 or spring-upgrade-plan/2 consumer implementations, review-only findings, standalone target selection or staged upgrade planning, runtime diagnosis, threat models, test-gap backlogs, or Modulith-only audits.

Use this skill when the user asks for an evidence-ranked static engineering review of a Spring or Spring Boot repository, code, configuration, dependencies, architecture, security controls, data or messaging design, production readiness, existing migration claims, or static capacity design against engineering best practices. Do not use it to implement changes, create upgrade plans, diagnose runtime artifacts, build threat models, plan test backlogs, verify Modulith graphs, or answer ordinary Spring questions.

Use this skill when the user asks to audit Spring application module boundaries, dependency cycles, internal API leakage, allowed dependencies, events, module tests, or module observability. Use it for an explicit dependency graph and Spring Modulith verification semantics. Do not add Modulith or change packages; hand requested implementation to spring-application-developer. Do not claim executable verification unless authorized execution is requested.

Use this skill when the user asks to map Spring production risks or change surfaces to missing and unverified tests across unit, slice, integration, contract, architecture, concurrency, migration, resilience, or performance levels. Use it to create a risk-ranked backlog with fixtures, oracles, CI placement, and flakiness controls. Do not use it to fix one failing test or equate line coverage with assurance.

Use this skill when the user asks for supported Spring Boot/Cloud, Java/Kotlin, Maven, or Gradle target selection or compatibility verification as the primary output; a staged, reversible migration plan; or implementation or validation of a spring-upgrade-plan/2 consumer. Use it for compatibility gates, version-hop sequencing, rollout, and rollback. Do not use it merely to review or apply an existing migration, or when version verification is incidental to a requested greenfield implementation. Bundled scripts require Python 3.12+.

Use this skill when the user asks for causal diagnosis of Spring/JVM latency, throughput, saturation, memory, GC, threads, pools, databases, messaging, or containers from runtime measurements and controlled experiments. Use it to rank hypotheses and confirm bottlenecks from JFR, metrics, traces, profiles, logs, or workload evidence. Without runtime artifacts, return a measurement plan; do not claim a bottleneck from static configuration alone.

Use this skill when the user asks to build an evidence-backed threat model for Spring HTTP, reactive, messaging, data, management, or outbound-integration surfaces. Use it to map assets, actors, trust boundaries, abuse paths, controls, residual risk, and testable acceptance criteria. Do not use it for active penetration testing, a CVE lookup, or legal/compliance advice.

Use only to initialize, refresh, audit, trim, reconcile, clean, or route active AGENTS.md or coding-agent instructions from local evidence. Never use to review a pull request or code change, even when it changes AGENTS.md; to run the built-in Codex /init command; or for translation, grammar-only edits, release notes, summaries, ADRs, proposals, explanations, or other adjacent deliverables.

Use only to initialize, refresh, audit, trim, reconcile, clean, or route active AGENTS.md or coding-agent instructions from local evidence. Never use to review a pull request or code change, even when it changes AGENTS.md; to run the built-in Codex /init command; or for translation, grammar-only edits, release notes, summaries, ADRs, proposals, explanations, or other adjacent deliverables.

Inspect, catalog, call, or safely refuse unofficial read-only 네이버 증권/네이버증권/네이버페이 증권/Npay 증권/Npay증권/npay증권/Naver Stock stock.naver.com web APIs for Korean and foreign stocks, US ETFs, indices, crypto, integrated search, home feeds, news, research, discussions, rankings, IPOs, charts, exchange rates, and market indicators. Use for stock.naver.com API discovery or data requests, including unsafe or non-read-only requests that must be refused or redirected.

Use this skill when the user asks for public, read-only TossInvest/토스증권 data visible on tossinvest.com, including Korean/US stock quotes, order books, candles, financials, filings, market search, news, rankings, screeners, calendars, industries/sectors, indices, FX, exchange-rate widgets, crypto-like index pages, sanitized public stock/lounge community data, an unofficial WebSocket API reference for browser-observed real-time trade/price streams, or public endpoint re-verification. Do not use for login, accounts, holdings, orders, authenticated broker workflows, bulk scraping, or investment advice.

Record, review, or update WATCHLIST.md/WL-YYYYMMDD-NNN deferred checks and 후속 체크; not generic reminders, wakeups, polling, or unscoped lifecycle requests.

Use only when explicitly invoked as briefprint, $briefprint, or /briefprint for cached document briefings: summarize, brief, digest, recap, or rerender documents, notes, tickets, logs, reports, transcripts, JSON/XML/API payloads, or cached briefing state. Do not use for ordinary one-off summaries, live research, source-code review/debugging, general writing, translation-only, simple Q&A, or non-cacheable input.

Design, implement, or review API/service reliability changes involving authentication or authorization, data integrity, idempotency, concurrency, external dependencies, migrations, compatibility, performance, or operational risk. Skip UI-only, metadata-only, trivial refactor, and prose-only work that cannot change API/service behavior, contracts, or operational risk.

Use only when the user explicitly invokes "$spring-best-practice-skill", selects it from /skills, or uses a runtime-specific direct skill invocation. Reviews Spring and Spring Boot architecture, code, dependencies, migrations, production readiness, security, Redis/cache, scheduling, data, messaging, batch, AI, and API design. Do not use for ordinary Spring questions.

Recoverable coding-session checkpoint at .savepoint/SAVEPOINT.md for context reset/session transfer. Use for 세이브포인트 만들어줘, 세이브포인트 로드해줘, 세이브포인트 읽어줘, 세이브포인트 이어서 해줘. Not for SQL SAVEPOINT, ordinary summaries, direct code/docs edits without checkpoint intent, /status, /new, PTY/session control, session rotation, or app features named savepoint.

Use when reviewing SOPs, PRDs, runbooks, tickets, handoffs, prompts, source-of-truth docs, or attached document bundles for execution-blocking ambiguity so a context-poor human, lightweight model, or agent can follow without guessing.

Use for unfamiliar code, new APIs, papers, prototypes, vague features, or root-cause debugging. Create a small verified artifact before full implementation.