diff --git a/.github/workflows/publish-alpha.yml b/.github/workflows/publish-alpha.yml
index 0dc1c3f..f2751be 100644
--- a/.github/workflows/publish-alpha.yml
+++ b/.github/workflows/publish-alpha.yml
@@ -28,6 +28,8 @@ jobs:
           node-version: 22
           cache: pnpm
           registry-url: https://registry.npmjs.org
+      - name: Install npm with trusted publishing support
+        run: npm install --global npm@latest
       - name: Install dependencies
         run: pnpm install --frozen-lockfile
       - name: Verify alpha package
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3d28628..ae452c4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,7 +2,9 @@
 
 ## Unreleased
 
-- No unreleased changes.
+- Use QVAC CLI's `127.0.0.1:11434` endpoint as the CLI and documentation default.
+- Add npm discovery keywords and a benefit-oriented package description.
+- Clarify that this is an independent community project.
 
 ## 0.1.0-alpha.1 - 2026-07-10
 
diff --git a/README.md b/README.md
index 0ff0f62..825b4c1 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,9 @@
 # qvac-bench
 
-Public QVAC developer tooling from localhost41.
+Command-line benchmarks for local QVAC OpenAI-compatible streaming endpoints.
+
+> Independent community project maintained by localhost41. Not affiliated with
+> or endorsed by Tether or QVAC.
 
 ## Local development
 
@@ -46,7 +49,7 @@ Benchmark a local QVAC OpenAI-compatible streaming endpoint:
 
 ```bash
 qvac-bench \
-  --url http://localhost:8000/v1/chat/completions \
+  --url http://127.0.0.1:11434/v1/chat/completions \
   --model qvac \
   --prompt "Say hello in one short sentence." \
   --max-tokens 64 \
@@ -182,7 +185,7 @@ pnpm build
 
 # 3. Warm the endpoint once without recording the result.
 node dist/cli.js \
-  --url http://localhost:8000/v1/chat/completions \
+  --url http://127.0.0.1:11434/v1/chat/completions \
   --model qvac \
   --prompt-name hello \
   --max-tokens 64 \
@@ -192,7 +195,7 @@ node dist/cli.js \
 
 # 4. Run the measured benchmark with the same inputs.
 node dist/cli.js \
-  --url http://localhost:8000/v1/chat/completions \
+  --url http://127.0.0.1:11434/v1/chat/completions \
   --model qvac \
   --prompt-name hello \
   --max-tokens 64 \
@@ -233,7 +236,7 @@ To reproduce a result:
 
    ```bash
    node dist/cli.js \
-     --url http://localhost:8000/v1/chat/completions \
+     --url http://127.0.0.1:11434/v1/chat/completions \
      --model qvac \
      --prompt-name hello \
      --max-tokens 64 \
diff --git a/package.json b/package.json
index 690c234..6fe67da 100644
--- a/package.json
+++ b/package.json
@@ -1,7 +1,16 @@
 {
   "name": "@localhost41/qvac-bench",
   "version": "0.1.0-alpha.1",
-  "description": "QVAC developer tooling package: qvac-bench",
+  "description": "Benchmark local QVAC OpenAI-compatible streaming endpoints from the command line",
+  "keywords": [
+    "qvac",
+    "benchmark",
+    "cli",
+    "llm-benchmark",
+    "local-ai",
+    "openai-compatible",
+    "performance"
+  ],
   "license": "MIT",
   "repository": {
     "type": "git",
diff --git a/src/cli.ts b/src/cli.ts
index 22eab82..ff081ea 100644
--- a/src/cli.ts
+++ b/src/cli.ts
@@ -11,7 +11,7 @@ QVAC benchmark developer CLI.
 Options:
   -h, --help              Show this help text
   --url <url>             QVAC OpenAI-compatible chat completions URL
-                          Default: http://localhost:8000/v1/chat/completions
+                          Default: http://127.0.0.1:11434/v1/chat/completions
   --model <model>         Model name to request. Default: qvac
   --prompt <prompt>       Prompt to send. Default: Say hello in one short sentence.
   --prompt-name <name>    Built-in prompt fixture to run: ${promptNames().join(", ")}
@@ -75,7 +75,7 @@ export interface BenchmarkDependencies {
 }
 
 const defaultOptions: BenchmarkOptions = {
-  url: "http://localhost:8000/v1/chat/completions",
+  url: "http://127.0.0.1:11434/v1/chat/completions",
   model: "qvac",
   prompt: "Say hello in one short sentence.",
   maxTokens: 64,
diff --git a/test/index.test.ts b/test/index.test.ts
index 489fb62..f0a985c 100644
--- a/test/index.test.ts
+++ b/test/index.test.ts
@@ -554,7 +554,7 @@ describe("qvac-bench", () => {
 
     const result = await measureQvacLatency(
       {
-        url: "http://localhost:8000/v1/chat/completions",
+        url: "http://127.0.0.1:11434/v1/chat/completions",
         model: "qvac-local",
         prompt: "hello",
         maxTokens: 8,