Design, review, or explain a new product's Go backend/microservice architecture using Kitex or similar RPC, Hertz or similar HTTP gateway, protobuf IDL, MySQL/GORM storage, Redis cache/locks/limits, service discovery, dynamic config, message queues, observability, DI, and code generation. Also owns multi-tenant isolation, event-driven/Kafka, and data-platform (sharding/replicas/HA) architecture. Product-agnostic; no dependence on existing codebase paths, service names, or legacy repos. Prefer this for architecture/boundary/contract/data-ownership/reliability/security/platform decisions with no code changes; use go-microservice-dev for implementation. Triggers include "Go 后端架构怎么设计", "Go 微服务怎么拆", "RPC 接口怎么定义", "Go 服务边界", "Go 服务重构 / 架构分层重构", "拆分 Go 服务的上帝类/模块边界 / decompose a god class".
Instrucciones de origen · Vista previa de solo lectura
name
go-microservice-architecture
description
Design, review, or explain a new product's Go backend/microservice architecture using Kitex or similar RPC, Hertz or similar HTTP gateway, protobuf IDL, MySQL/GORM storage, Redis cache/locks/limits, service discovery, dynamic config, message queues, observability, DI, and code generation. Also owns multi-tenant isolation, event-driven/Kafka, and data-platform (sharding/replicas/HA) architecture. Product-agnostic; no dependence on existing codebase paths, service names, or legacy repos. Prefer this for architecture/boundary/contract/data-ownership/reliability/security/platform decisions with no code changes; use go-microservice-dev for implementation. Triggers include "Go 后端架构怎么设计", "Go 微服务怎么拆", "RPC 接口怎么定义", "Go 服务边界", "Go 服务重构 / 架构分层重构", "拆分 Go 服务的上帝类/模块边界 / decompose a god class".
Go Microservice Architecture
Use this for new product/server architecture work. It distills proven backend patterns but must not assume any existing repository, service identifier, package path, database, or legacy service layout.
Skill Routing
Use this skill for architecture decisions, service decomposition, API/RPC contract shape, storage ownership, async workflow design, observability, reliability, and launch readiness.
Use product-rd-workflow first when the request spans product shaping, architecture, implementation, review, release, and learning loops rather than architecture alone.
Use go-microservice-dev when the user asks to implement, modify, debug, test, scaffold, or generate code. A localized refactor (某文件/某类) also belongs to go-microservice-dev; only service-wide layering/boundary redesign stays here, and a multi-stage / cross-module refactor delivery must re-enter product-rd-workflow. An unqualified whole-service "Go 服务重构" enters here by default, mirroring the Python pair — do not let the unqualified form fall to the implementation skill.
Use python-service-architecture for Python services. Do not load Python rules for Go work unless the task is explicitly cross-language contract design.
Use platform-observability for logs/metrics/traces/log-id propagation/dashboards/alerts/SLI-SLO design. This skill owns only the service-side observability surface (Kitex/Hertz middleware that emits structured logs, OTel SDK use, baseline labels attached by metrics client, health endpoints); the cross-cutting evidence stack belongs there.
Use platform-service-connectivity for service mesh, service discovery (Nacos/Consul/equivalent), mTLS, multi-environment lane routing, retry/timeout/circuit-breaker policy, and the canonical RPC base struct shape. This skill defines what the service exposes (framework default options usage, request-metadata propagation conformance via the platform's chosen carrier — in-message base or transport metadata, scenario-driven per the connectivity Carrier decision — graceful shutdown); mesh policy and routing rules live there.
Use platform-release-engineering for environment/lane matrix, canary/blue-green, promotion gates, rollback playbook, secret distribution, dynamic-config (config-center) vs static-config split, image build pipeline. This skill owns the service-side contracts (config schema, secret consumption interface, registration metadata); release flow lives there.
Use codebase-specific skills only when the task is explicitly about an existing legacy/workspace repository.
Architecture references should define decision criteria, invariants, ownership boundaries, and acceptance checks. Put concrete code patterns, helper APIs, and test mechanics in go-microservice-dev.
When changing an architecture rule that implementation must obey, name the downstream execution owner before landing: go-microservice-dev for code mechanics, testing-strategy for proof layer, platform skills for runtime contracts, and product-rd-workflow for cross-stage gates. If the rule also applies outside Go, route through skill-extraction-workflow and mirror or explicitly skip sibling architecture/dev skills.
For money, billing, quota, permission, tenant/user data isolation, privacy, high-impact AI, repeated writes, async finality, or incident-explanation risk, start from product-rd-workflow and its high-risk resilience gate before choosing service boundaries or fallback behavior.
Platform Boundary
This skill owns the service-internal view: layer separation (transport/application/domain/infra), protobuf IDL design, DAL boundaries, MQ consumer shape, error-code contract, DI/Wire generation. It exposes a fixed contract to the platform layer:
A Go service author works against this contract. Platform-level questions ("which mesh policy?", "which collector?", "which canary weight?") route to the platform skills. Cross-language uniformity is what makes the platform layer reusable.
Generalization Discipline
Keep only reusable server-side mechanics: contracts, layers, storage ownership, config, reliability, observability, security boundaries, code generation, and delivery workflow.
Do not copy product nouns, service names, package paths, provider names, environment names, IDs, dashboards, callback types, or organization-specific operating habits into the architecture.
When source code shows a domain-specific concept, generalize it to domain object, resource, authorization scope, integration scope, or workflow only if the mechanic is reusable; otherwise discard it.
When patterns conflict, choose deliberately:
Prefer current contracts and requirements over historical code habits.
Prefer relational durable truth over Redis-only truth for auditable state.
Prefer explicit protobuf/config/schema contracts over reflection, hidden globals, or stringly conventions.
Prefer idempotency and durable acceptance over retry-only designs.
Prefer fail-closed for auth, permission, critical state transitions, and data-integrity paths; fail-open only for non-critical cache/telemetry paths when availability requires it.
Fuse patterns only when both are generic and complementary; otherwise keep the simpler product-agnostic rule.
Example: if one codebase stores workflow state only in Redis and another persists it in a relational table, keep the relational source-of-truth rule and optionally use Redis as cache, lock, rate limit, or scheduling aid.
Core Workflow
Before changing architecture guidance, contracts, service boundaries, diagrams, plans, or implementation-driving recommendations, complete enough analysis and planning for the decision to be reviewable. Scale the plan to risk: a simple low-risk explanation can use a short inline plan; multi-service, contract-visible, data-ownership, release, high-risk, branch/MR, or unclear-risk architecture work needs explicit assumptions, alternatives, tradeoffs, acceptance checks, verification evidence, rollback or migration path, and named handoffs to implementation, testing, platform, or product workflow skills before edits or approval.
Define the product boundary.
Identify user workflows, domain objects, external integrations, and lifecycle events.
Decide which capabilities must be synchronous APIs, internal RPCs, async jobs, or scheduled tasks.
Internal microservice calls use RPC/gRPC by default and may use HTTP when the service contract chooses it. Internal HTTP must meet the same service discovery, auth, timeout, retry, observability, and contract-test standards as RPC.
Avoid splitting into microservices until ownership, scaling, data boundary, or deployment cadence justifies it.
Define service contracts before implementation.
Use protobuf IDL as the default contract format.
Service identity is decided before IDL and implementation. For new services, define the globally unique service name first, then IDL, then generated contract artifacts, then business implementation.
Treat IDL and idlgen outputs as a stable boundary between service contracts and business repositories. Prefer centralized IDL ownership plus versioned generated artifacts that business services import. When the platform uses shared IDL and IDLGen repositories, Go, Python, web, mobile, and mini-program consumers must derive from the same contract source; do not fork IDL or generated artifacts per service repository or per language unless it is a deliberate compatibility branch with owner, version, and deprecation plan. Service repository layout (single service, multi-service, API/RPC split) does not change contract ownership.
Protobuf may be the contract source for HTTP as well as RPC. For protobuf-backed or cross-language HTTP gateways, decide separately: contract source (protobuf/OpenAPI), generated route/client artifacts, and wire format (JSON vs binary protobuf). Classify protobuf-backed HTTP using the canonical wire-format gate in ../platform-service-connectivity/references/protobuf-http-contract-signals.md; do not restate or widen that predicate locally.
API contracts must define request messages, the response envelope, and shared public fields in the shared contract source before implementation, per the canonical code/message/data envelope contract in ../platform-service-connectivity/references/http-response-envelope-contract.md (adoption, migration, non-JSON surfaces, and anti-patterns live there — do not restate them here).
Internal RPC contracts carry the platform's shared request/response metadata field when the platform defines one; platform-service-connectivity owns the canonical base-struct shape and named-framework examples. Do not inherit a framework's reserved envelope/base field number as a cross-product convention (per references/protobuf-contract-architecture.md); if a shared metadata field must be part of the protobuf contract (rather than carried via middleware metadata), a new product picks an explicit field and documents it. Within a platform that has already standardized such a field, its field number, name, and message type are compatibility surfaces — do not renumber, rename, or replace them locally.
Architecture Defaults
One product can start as a modular monolith if service boundaries are unclear; split by data ownership and team/runtime needs, not by noun count.
Public API layer should be thin: auth, validation, request/response mapping, orchestration handoff.
RPC services should own domain operations and internal contracts.
Background workers should own event processing, retries, idempotency, and backfill/repair workflows.
A mature Go service usually separates transport/handler, application logic, domain/service orchestration, and infrastructure adapters; dependency injection or provider sets should make those seams visible instead of letting handlers construct DB/MQ/RPC clients directly.
Shared libraries are allowed for infrastructure concerns; do not put domain rules in shared packages.
Cross-boundary semantic values are an architecture responsibility even when their local representation is small. Architecture decides where finite values such as status, market, region, channel, source, provider, or permission are canonical, who may expose generated transport enums to domain code, and how duplicate constants are retired.
Mature reference code is useful only when it exposes a reusable boundary. Prefer extracting framework wrappers, typed clients, context/error contracts, config schemas, transaction helpers, MQ lifecycle helpers, and query-safety checks; discard business nouns, private module layout, and one-off legacy habits.
A service should have clear layers: transport, application/service logic, domain logic, infrastructure adapters, generated contracts.
Long-running jobs require a first-class runtime model: distributed lock scope, lease time, max execution time, retry policy, idempotency key, and failure visibility.
External dependencies require explicit timeout budgets, status-code validation, retry/fallback rules, and clear ownership of degraded behavior.
High-risk operations require a resilience contract: fail-closed policy, idempotency strategy, durable status, reconciliation or repair path, trace/request id propagation, user/support explanation surface, and proof that fallback/degradation cannot bypass authorization, tenant/user isolation, quota, audit, or data-retention controls.
High-risk context resolution must reject missing tenant, actor, subject, or resource scope instead of falling back to default identities. Durable side effects need atomic audit/outbox evidence or an explicit reconciliation/repair workflow.
External dependency bootstrap requires a first-class operability contract: which resources may be created automatically, which existing resources must be validated, which partial failures can be repaired, which operations fail startup, and which timeouts apply to administrative work versus serving requests.
Reference Loading
For source provenance, current extraction boundary, and keep/merge/discard decisions, read references/source-evidence-map.md when auditing or re-extracting this skill.
For architecture decisions and boundaries, read references/architecture-playbook.md.
For reliability and operability requirements, read references/ops-checklist.md.
For cross-cutting server contracts, read references/cross-cutting-concerns.md.
For service scaffolding, generation, DI, config, and test architecture, read references/service-scaffold.md.
For protobuf/IDL contract design, package boundaries, message/enum/service evolution, and HTTP/RPC mapping, read references/protobuf-contract-architecture.md.
For secret providers, service discovery, dependency clients, object storage, and controlled concurrency, read references/dependency-platform.md.
For public API, third-party callbacks, authorization scope isolation, and external integration boundaries, read references/api-security-boundaries.md.
For HTTP gateway boundaries, generated routes, handler responsibilities, middleware order, and generated HTTP clients, read references/http-gateway-architecture.md.
For deployment, environment/lane runtime, mesh/gateway traffic, canary, approval, rollback, and launch operations, read references/release-runtime-readiness.md.
For runtime identity, environment/lane metadata, service discovery, logs, metrics, tracing, and debug exposure, read references/runtime-observability.md.
For relational data modeling, schema ownership, indexes, sharding, migrations, and generated DAL boundaries, read references/data-modeling-and-migrations.md.
For Redis cache, locks, counters, rate limiting, task coordination, local cache, and idempotency architecture, read references/redis-cache-coordination.md.
For import/export, backfill, migration, partitioning, artifact policy, and batch limits, read references/bulk-workflow-architecture.md.
For durable workflow state, allowed transitions, retry, cancellation, idempotency, and terminal-state policy, read references/workflow-state-architecture.md.
For MQ topics, consumer groups, event contracts, activation gates, retry/drop policy, and async consumer operations, read references/mq-consumer-architecture.md.
For event-driven architecture concerns — delivery semantics taxonomy, producer-side patterns, transactional outbox/inbox, idempotency design, partition-key ordering, schema evolution, retry/DLQ/replay strategy, fanout patterns, saga vs choreography, end-to-end "exactly-once" illusion, and Go-specific implementation glue (consumer goroutine + worker pool, outbox poller with , error classification, graceful-shutdown order) — read . Stack-agnostic core sections mirror the sibling ; maintainers updating those sections must update both files in the same change.
Internal service data models need the same contract discipline: shared cross-service DTOs, enums, status values, metadata fields, and request/response models live in the contract or generated artifact boundary; service-private domain and persistence models stay inside the owning service and convert at transport/application boundaries.
Separate public HTTP API models from internal RPC models when external clients have different stability needs.
Prefer additive contract evolution: new fields, new methods, new enum values; avoid breaking field numbers or response semantics.
Inventory the service's exposed surfaces and review new ones at the boundary. Keep a machine-checkable inventory of every externally reachable surface appropriate to the stack — HTTP routes, RPC services and methods, and async subscriptions (MQ / event / webhook consumers); a surface absent from it should not reach production unreviewed. Drive it from a deterministic discovery profile (routes/subscriptions-as-code, or a generated manifest plus a checked-in snapshot, are two common patterns) with explicit allowlisted carve-outs for framework / health / debug, generated, plugin, and env-conditional surfaces, so the gate flags genuinely new exposure instead of churning on false positives; also flag inventory entries no longer present in code so the snapshot stays trustworthy. Scale enforcement to risk: hard-fail CI for production, externally reachable services; a lighter manifest + review checklist suffices for prototypes, internal scripts, or repos without CI. The inventory proves every surface was seen and reviewed, not that it is authorized — it is not an authn/authz gate; auth, tenant isolation, and safe exposure remain separate evidence the boundary review must still demand. An unregistered new route or consumer is a shadow surface (unreviewed; async consumers are a commonly missed class). Distinct from breaking-change detection (buf breaking / kitex check, which guards changes to existing surfaces) and from agent-contract directory coverage (whether a directory carries an AGENTS.md); it guards new runtime exposure. product-rd-workflow's spec/repo-contract sync gate owns the human discipline; this is its mechanical enforcement. Mirror any change to this rule in ../python-service-architecture/SKILL.md.
For finite values that cross service, storage, client, analytics, or generated-code boundaries, architecture must name the canonical owner, shared package or contract location, conversion boundaries, unknown/default behavior, and migration/debt exit path before implementation. If no shared location exists, approve the local-slice fallback and require every finite-value-debt marker to carry task reference, owner, deadline, and reason.
Design data ownership.
Each service owns its write model and schema.
Cross-service reads should go through RPC/API or explicit read models, not shared table writes.
Use relational DB as source of truth for transactional domain state.
Use GORM as the default MySQL DAL choice for new Go services unless the repo already standardizes or conventionally uses another query layer, or the architecture records a deliberate alternative.
Decide schema, indexes, unique constraints, migration path, and sharding keys before writing DAL code.
Keep GORM, sqlx, and raw SQL behind DAL or repository boundaries; transport and application layers must not build database queries directly.
Use Redis for cache, locks, counters, idempotency windows, rate limits, and ephemeral coordination, not durable truth.
Define Redis key scope, TTL policy, lock lease, rate-limit window, and idempotency lifetime as part of the architecture.
Design runtime platform contracts.
Config: local file + dynamic config, with environment/lane separation.
High-risk feature flags and runtime config default fail-closed in production. Local stubs, test defaults, and developer-safe switches must be visibly scoped and cannot become implicit production enablement.
Secrets: never stored in code or config files; resolve through a secret provider.
Service discovery: all internal RPC clients use discovery/resolver abstraction.
Service identity: internal RPC/HTTP trust should be based on workload identity, mTLS, signed service tokens, or an equivalent zero-trust control, not only network location.
Runtime wrappers should provide service identity, registration/discovery, context propagation, trace/log id, metrics, timeout/circuit behavior, and safe debug exposure consistently across HTTP, RPC, and workers.
A capability requirement is not a framework-adoption decision. When a requirement names a capability (service discovery, mTLS, graceful drain), first define its minimal closure — the smallest layer that satisfies it — before adopting a platform framework or runtime component for it, AND name which layer actually closes it: service discovery and mTLS can close at the deployment layer (orchestrator DNS, mesh sidecar), but graceful drain inherently needs runtime behavior (stop accepting, SIGTERM/readiness handling, in-flight completion) — manifests alone leave it explicitly not closed, never silently satisfied. Framework adoption is not all-or-nothing: when a framework default conflicts with a service characteristic (e.g. a fixed 30s drain window versus minute-scale streaming requests), split at the boundary — land the part that delivers value on its own (deployment manifests), decide or defer the conflicting part (runtime lifecycle adoption) separately with each deferred capability's closure status recorded, and keep a quotable record of why it was not adopted. Mirror any change to this rule in ../python-service-architecture/SKILL.md.
Dependency clients: DB, Redis, MQ, object storage, service discovery, secret provider, and external HTTP clients must have explicit timeout, credential, observability, and test-substitution contracts.
Dependency platform layers own SDK lifecycle, credential resolution, readiness, shared timeout policy, close hooks, and test substitution. Feature infrastructure adapters own feature-specific schemas, indexes, query/ranking contracts, object names, event payload meaning, and repair semantics. Do not place feature contracts in the platform layer just because they use a platform client.
Admission control: inbound rate limits, concurrency limits, load shedding, backpressure, and circuit-breaker behavior are part of reliability design.
Release runtime: service metadata, deployment resources, mesh/gateway routing, canary traffic, approval, and rollback are first-class architecture surfaces.
Error contract: define canonical code/message semantics once, then map HTTP, RPC, worker, and dependency errors into it.
Context contract: trace/log id, lane/environment, authorization/resource scope, request deadline, and auth subject must propagate through all sync and async paths.
Observability: logs, metrics, traces, health checks, pprof/debug where safe, and structured request context.
When comparing a new product backend against a mature internal reference, convert the comparison into a platform-capability gap list, not a source-code shopping list.
The recurring maturity checks are: identity/RBAC subject model, external API signature/replay/allowlist boundary, dynamic config with version/cache/watch/rollback, MQ producer/consumer lifecycle with retry/drop/replay visibility, DB transaction and query-safety guardrails, context/error propagation across HTTP/RPC/MQ, and resource/search visibility scope.
A placeholder or README-only package family is not implementation evidence.
Before marking an internal reference package family as placeholder-only, inspect nested git repositories or submodules, non-default local/remote branches, tags, and tree contents with read-only commands. Default-branch scaffolds do not prove the architecture capability is absent.
If the comparison uses internal checkouts, private repositories, local paths, or organization projects, route preservation through skill-extraction-workflow or apply the same sanitization gate before landing any artifact: keep only mechanisms, boundaries, and acceptance gates; remove source-identifying domains, paths, repository/module names, people, tickets, and business nouns.
Define code generation and ownership boundaries.
Generated code is an output surface, not hand-maintained source.
Keep IDL, generated contracts, HTTP route generation, DB model generation, and Wire/DI generation explicit.
Introducing or migrating any HTTP gateway/framework requires a quotable record in the architecture doc, repo contract, or MR description, with a separately locatable entry for gateway-capability justification, contract source, middleware order, generated artifacts, observability, and tests. Protobuf motivation alone is insufficient evidence unless the same record states a concrete missing capability in the existing framework, why the current extension points cannot satisfy it, and which requirement the new framework satisfies. Adding protobuf-backed HTTP on an unchanged gateway/framework requires the contract-source and wire-format records; expand to generated-artifact, observability, and test records when those surfaces change.
Health checks should be able to prove the boundary: IDL source location, generation command, generated package version, service implementation importing generated contracts, and CI/breaking-change checks. Agent review then judges whether the service layer bypasses the contract or hides domain semantics in transport code.
Document the generation commands in the repo.
Generation commands must be portable and must not contain developer-local absolute paths.
