Return explicit errors from AlertAPI construction instead of panicking inside the library layer, and let main own process-level failure handling.
Also delete the unused config.MustLoad helper and lock the supported config loading paths with tests.
Reduce the domain outbox surface to the shared retry/backoff contract that is actually consumed by the runtime layer.
Also make batch compensation processing pick up retrying records explicitly and cover that path with tests.
Use a shared in-memory code store across mock, Tencent, and Aliyun SMS services so send and verify follow the same contract. Also surface batch flush failures through FlushNow and explicit error tracking hooks for audit buffering.
Add broader e2e coverage for account, package, billing, tracing, and reliability scenarios.\nSupport Unix socket DSN formatting in config and cover it with unit tests.\nIgnore local assistant metadata and generated gate artifacts to reduce workspace noise.
1. 添加 slog_logger.go 实现基于 Go 1.21+ slog 的结构化日志
2. 支持 trace_id、request_id、tenant_id 等标准字段注入
3. 添加日志标准化重构方案文档
推荐使用 Go 内置 log/slog，无需第三方依赖。
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add regression tests for the shipped development and SMS sample configs, and fix the SMS example to match the runtime flat Config schema instead of nested provider blocks. Verified with fresh go test runs for ./internal/config and ./internal/sms before commit.
Add repository integration probes, repository policy tests, the compose-based integration runner, and the matching usage documentation. Align the runner environment with both repository and middleware integration test expectations, and verify with fresh repository tests, integration-tag test runs, bash -n, and docker-compose config before commit.
Add benchmark documentation and middleware benchmark coverage, fix the settlement benchmark mock to satisfy the current SettlementStore interface, and add reusable domain test helper packages. Verified with fresh go test runs for ./internal/testutil/... and go test -tags=slow -run '^$' ./internal/benchmark/... before commit.
Add the domain-level outbox processor implementation required by existing domain tests, and add regression coverage for compensation worker cancellation behavior. Verified with a fresh go test run for ./internal/domain before commit.
Add the outbox, partitioning, and token-status DDL files alongside the partition strategy regression test. These files map directly to already committed repository and middleware paths, and were verified with fresh repository, outbox, and middleware test runs before commit.
Add the untracked SMS service package and its security-focused tests, plus database-backed IAM service unit coverage. This keeps the verification surface small and brings previously validated test/runtime support sources into version control as an isolated change.
Check in the healthcheck, structured logging, outbox broker, partition manager, and token status repository files that the committed supply-api runtime already imports. Verified with fresh go test runs for cmd/supply-api, internal/httpapi, internal/pkg/logging, internal/repository, and internal/outbox.
Add nil and wildcard coverage for IAM claims helpers, ensure auth security tests run with authentication enabled, and make settlement mocks express pending-withdraw/error branches explicitly.
- audit/events: 73.5% → 97.6% (+24.1%)
- Add tests for IsM013/M014/M015RelatedEvent
- Add tests for FormatSECURITYEvent
- Add comprehensive coverage for all CRED and SECURITY event functions
- security: 67.2% → 88.8% (+21.6%)
- Add tests for ValidateKeyID, DecryptionError.Error()
- Add tests for ValidateQueryParams, GetAllowedParamNames
- Add tests for isHexString, looksLikeAPIKey
- Fix test cases to match actual implementation behavior
- audit/sanitizer: Fix MaskMap []string handling bug
- Add maskSliceInterface for []interface{} type
- Tests now pass for string slice sensitive fields
All tests pass
- Fix MaskMap to properly handle []string sensitive fields
- Add missing slice handling in sanitizer
- Add comprehensive tests for GetMetrics and CreateEventsBatch
- Improve audit/handler coverage from 49.8% to 68.8%
- Fix test expectations to match actual sanitizer behavior
- All tests pass
