spencercjh
diff --git a/‎.gitignore‎
Lines changed: 3 additions & 0 deletions b/‎.gitignore‎
Lines changed: 3 additions & 0 deletions
diff --git a/‎CLAUDE.md‎
Lines changed: 19 additions & 1 deletion b/‎CLAUDE.md‎
Lines changed: 19 additions & 1 deletion
diff --git a/‎cmd/enrich.go‎
Lines changed: 12 additions & 3 deletions b/‎cmd/enrich.go‎
Lines changed: 12 additions & 3 deletions
diff --git a/‎…rs/plans/2026-03-27-gozero-unit-tests.md‎ ‎…cs/plans/2026-03-27-gozero-unit-tests.md‎docs/superpowers/plans/2026-03-27-gozero-unit-tests.md renamed to docs/plans/2026-03-27-gozero-unit-tests.md b/‎…rs/plans/2026-03-27-gozero-unit-tests.md‎ ‎…cs/plans/2026-03-27-gozero-unit-tests.md‎docs/superpowers/plans/2026-03-27-gozero-unit-tests.md renamed to docs/plans/2026-03-27-gozero-unit-tests.md
@@ -43,6 +43,9 @@ Thumbs.db
 .spec-forge.yaml
 !.spec-forge.example.yaml
 
+# E2E test local config (in integration-tests/)
+integration-tests/.spec-forge.e2e.local.yaml
+
 # Environment files
 .env
 .env.local
 
@@ -27,6 +27,10 @@ make fmt
 make verify
 ```
 
+> **Important:** `make verify` checks for uncommitted changes (git diff) and will fail if there are pending changes.
+> Before committing code, use individual commands: `make fmt`, `make lint`, `make test`.
+> Only use `make verify` after committing or in CI environments where working tree is clean.
+
 ## Architecture Overview
 
 Spec Forge is a CLI tool that generates enriched OpenAPI specifications from various frameworks (Spring Boot, go-zero, gRPC-protoc).
@@ -239,16 +243,30 @@ Test schema field and API parameter enrichment with DeepSeek:
 # Build the binary first
 go build -o ./build/spec-forge .
 
-# Run enrichment with Chinese descriptions
+# Run enrichment with Chinese descriptions (streaming enabled by default)
 LLM_API_KEY="your-deepseek-api-key" ./build/spec-forge enrich \
     ./integration-tests/maven-springboot-openapi-demo/target/openapi.json \
     --provider custom \
     --model deepseek-chat \
     --custom-base-url https://api.deepseek.com/v1 \
     --language zh \
     -v
+
+# Or use --no-stream for faster processing (enables concurrent LLM calls)
+LLM_API_KEY="your-deepseek-api-key" ./build/spec-forge enrich \
+    ./integration-tests/maven-springboot-openapi-demo/target/openapi.json \
+    --provider custom \
+    --model deepseek-chat \
+    --custom-base-url https://api.deepseek.com/v1 \
+    --language zh \
+    -v \
+    --no-stream
 ```
 
+> **Note:** Streaming is enabled by default, showing real-time LLM output with batch-type prefixes
+> (`[api]`, `[schema]`, `[param]`). With streaming on, batches are processed sequentially for readable output.
+> Use `--no-stream` to enable concurrent processing across batches for faster enrichment.
+
 Expected output:
 - Schema fields get Chinese descriptions (e.g., `User.id` → "用户唯一标识符")
 - API parameters get Chinese descriptions (e.g., `page` → "指定要获取的页码，用于分页查询")
 
@@ -67,6 +67,8 @@ func runEnrich(cmd *cobra.Command, args []string) error {
 	customBaseURLFlag, _ := cmd.Flags().GetString("custom-base-url")
 	//nolint:errcheck
 	customAPIKeyEnvFlag, _ := cmd.Flags().GetString("custom-api-key-env")
+	//nolint:errcheck
+	noStreamFlag, _ := cmd.Flags().GetBool("no-stream")
 
 	// Determine provider
 	prov := providerFlag
@@ -153,7 +155,11 @@ func runEnrich(cmd *cobra.Command, args []string) error {
 	}
 
 	// Enrich
-	result, err := e.Enrich(ctx, spec, &enricher.EnrichOptions{Language: lang})
+	streamEnabled := !noStreamFlag // Streaming enabled by default
+	result, err := e.Enrich(ctx, spec, &enricher.EnrichOptions{
+		Language: lang,
+		Stream:   &streamEnabled,
+	})
 	if err != nil {
 		// Check if partial enrichment
 		if partialErr, ok := errors.AsType[*processor.PartialEnrichmentError](err); ok {
@@ -282,10 +288,11 @@ Examples:
 	c.Flags().String("model", "", "LLM model name")
 	c.Flags().String("language", "en", "Output language for descriptions")
 	c.Flags().StringP("output", "o", "", "Output file (default: overwrite input)")
-	c.Flags().Int("concurrency", 3, "Number of concurrent LLM calls")
+	c.Flags().Int("concurrency", 3, "Max concurrent LLM calls (only effective with --no-stream)")
 	c.Flags().Duration("timeout", 30*time.Second, "Timeout for single LLM call")
 	c.Flags().String("custom-base-url", "", "Custom provider API URL")
 	c.Flags().String("custom-api-key-env", "LLM_API_KEY", "Environment variable for custom API key")
+	c.Flags().Bool("no-stream", false, "Disable streaming to enable concurrent processing (faster, but no real-time output)")
 
 	return c
 }
@@ -300,6 +307,7 @@ var (
 	enrichTimeout         time.Duration
 	enrichCustomBaseURL   string
 	enrichCustomAPIKeyEnv string
+	enrichNoStream        bool
 )
 
 func init() {
@@ -309,8 +317,9 @@ func init() {
 	enrichCmd.Flags().StringVar(&enrichModel, "model", "", "LLM model name")
 	enrichCmd.Flags().StringVar(&enrichLanguage, "language", "en", "Output language for descriptions")
 	enrichCmd.Flags().StringVarP(&enrichOutput, "output", "o", "", "Output file (default: overwrite input)")
-	enrichCmd.Flags().IntVar(&enrichConcurrency, "concurrency", 3, "Number of concurrent LLM calls")
+	enrichCmd.Flags().IntVar(&enrichConcurrency, "concurrency", 3, "Max concurrent LLM calls (only with --no-stream)")
 	enrichCmd.Flags().DurationVar(&enrichTimeout, "timeout", 30*time.Second, "Timeout for single LLM call")
 	enrichCmd.Flags().StringVar(&enrichCustomBaseURL, "custom-base-url", "", "Custom provider API URL")
 	enrichCmd.Flags().StringVar(&enrichCustomAPIKeyEnv, "custom-api-key-env", "LLM_API_KEY", "Environment variable for custom API key")
+	enrichCmd.Flags().BoolVar(&enrichNoStream, "no-stream", false, "Disable streaming output to enable concurrent LLM calls (faster)")
 }