Fix CDN deployment docs and release packaging

Drop the unused HIVE_DATA KV requirement from self-host instructions, and package release zips with the existing action-zip path so runners do not need a system zip binary.

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: lisa-0831

.github/workflows/build-and-dockerize.yaml

@@ -101,7 +101,7 @@ jobs:
             packages/services/cdn-worker/dist/index.lambda.mjs
           dest: ${{ inputs.imageTag }}.zip
 
-      - name: archive CDN release artifacts
+      - name: prepare CDN release artifacts
         if:
           ${{ inputs.uploadJavaScriptArtifacts && inputs.publishLatest && github.ref_type == 'tag'
           && github.ref_name == format('hive@{0}', inputs.latestVersion) && matrix.platform ==
@@ -110,8 +110,30 @@ jobs:
           mkdir -p release-assets/cloudflare release-assets/lambda
           cp packages/services/cdn-worker/dist/index.worker.mjs release-assets/cloudflare/index.mjs
           cp packages/services/cdn-worker/dist/index.lambda.mjs release-assets/lambda/index.mjs
-          (cd release-assets/cloudflare && zip ../hive-cdn-cloudflare-worker.zip index.mjs)
-          (cd release-assets/lambda && zip ../hive-cdn-aws-lambda.zip index.mjs)
+
+      # recursive: true puts directory contents at the zip root (index.mjs), matching the
+      # Lambda handler `index.handler` and Wrangler `main = "index.mjs"`.
+      - uses: vimtor/action-zip@5f1c4aa587ea41db1110df6a99981dbe19cee310 # v1
+        name: archive CDN Cloudflare Worker release artifact
+        if:
+          ${{ inputs.uploadJavaScriptArtifacts && inputs.publishLatest && github.ref_type == 'tag'
+          && github.ref_name == format('hive@{0}', inputs.latestVersion) && matrix.platform ==
+          'linux/amd64' }}
+        with:
+          recursive: true
+          files: release-assets/cloudflare/
+          dest: release-assets/hive-cdn-cloudflare-worker.zip
+
+      - uses: vimtor/action-zip@5f1c4aa587ea41db1110df6a99981dbe19cee310 # v1
+        name: archive CDN AWS Lambda release artifact
+        if:
+          ${{ inputs.uploadJavaScriptArtifacts && inputs.publishLatest && github.ref_type == 'tag'
+          && github.ref_name == format('hive@{0}', inputs.latestVersion) && matrix.platform ==
+          'linux/amd64' }}
+        with:
+          recursive: true
+          files: release-assets/lambda/
+          dest: release-assets/hive-cdn-aws-lambda.zip
 
       - name: upload CDN release artifacts
         if:

packages/services/cdn-worker/README.md

@@ -66,26 +66,19 @@ curl --fail --location \
 unzip hive-cdn-cloudflare-worker.zip
 ```
 
-The extracted `index.mjs` is an ES module Worker. Create its required Workers KV namespace:
+The extracted `index.mjs` is an ES module Worker. Create a `wrangler.toml` next to it. The
+following is a minimal configuration; the six Analytics Engine bindings are required. Dataset
+names may be changed, but the binding names must remain unchanged.
 
-```sh
-npx wrangler kv namespace create HIVE_DATA
-```
-
-Copy the namespace ID printed by Wrangler into `wrangler.toml`. The following is a minimal
-configuration; the six Analytics Engine bindings are required. Dataset names may be changed, but the
-binding names must remain unchanged.
+The current Worker artifact does not read a Workers KV binding. Hive's Pulumi stack still creates a
+legacy `HIVE_DATA` namespace for historical reasons; you do not need it for a working deployment.
 
 ```toml
 name = "hive-cdn"
 main = "index.mjs"
 compatibility_date = "2026-03-03"
 workers_dev = true
 
-[[kv_namespaces]]
-binding = "HIVE_DATA"
-id = "<KV_NAMESPACE_ID>"
-
 [[analytics_engine_datasets]]
 binding = "USAGE_ANALYTICS"
 dataset = "hive_ha_cdn_usage_production"