From 45166d121896f2b13eaca5673f77e0b8f9d71721 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 14:52:47 +0000
Subject: [PATCH 01/10] fix: patch xml2js prototype pollution, route OpenCode
 review through hush
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Add npm override for xml2js ^0.6.2 (fixes Dependabot alert #1,
  prototype pollution in transitive dep blessed-contrib → map-canvas → xml2js)
- Integrate hush@0.1.7 into the OpenCode AI review workflow:
  install and start the hush gateway on :4000, copy the hush plugin,
  configure opencode.json to route API calls through the proxy.
  Defense-in-depth: plugin blocks sensitive file reads, proxy redacts
  PII from normal file content before it reaches the model.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 31 +++++++++++++++++++++++----
 package-lock.json                     | 10 ++++-----
 package.json                          |  3 +++
 3 files changed, 35 insertions(+), 9 deletions(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index b032395..1ca0308 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -69,16 +69,39 @@ jobs:
             echo "skip=false" >> $GITHUB_OUTPUT
           fi
 
+      - name: Setup Node.js
+        if: steps.check_changes.outputs.skip != 'true'
+        uses: actions/setup-node@v4
+        with:
+          node-version: 22
+
+      - name: Start Hush Gateway
+        if: steps.check_changes.outputs.skip != 'true'
+        run: |
+          npm install -g @aictrl/hush@0.1.7
+          PORT=4000 HUSH_HOST=127.0.0.1 hush &
+          # Wait for gateway to be ready
+          for i in $(seq 1 20); do
+            curl -sf http://127.0.0.1:4000/health > /dev/null 2>&1 && break
+            sleep 0.5
+          done
+          echo "Hush gateway running on :4000"
+
       - name: Setup OpenCode
         if: steps.check_changes.outputs.skip != 'true'
         env:
           ZHIPU_API_KEY: ${{ secrets.ZHIPUAI_API_KEY }}
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
-          # Use GITHUB_TOKEN to avoid rate limits when fetching version info
+          # Install OpenCode
           curl -fsSL https://opencode.ai/install | bash -s -- --no-modify-path
           echo "$HOME/.opencode/bin" >> $GITHUB_PATH
 
+          # Configure OpenCode to route through hush proxy + hush plugin
+          mkdir -p .opencode/plugins
+          cp examples/team-config/.opencode/plugins/hush.ts .opencode/plugins/hush.ts
+          printf '%s\n' '{"provider":{"zai-coding-plan":{"options":{"baseURL":"http://127.0.0.1:4000/api/coding/paas/v4"}}},"plugin":[".opencode/plugins/hush.ts"]}' > opencode.json
+
       - name: Direct OpenCode Review
         if: steps.check_changes.outputs.skip != 'true'
         env:
@@ -87,7 +110,7 @@ jobs:
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
           echo "Starting review with GLM-5 for SHA $SHA..."
-          
+
           $HOME/.opencode/bin/opencode run --model zai-coding-plan/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
 
           Focus areas:
@@ -95,7 +118,7 @@ jobs:
           2. **Streaming Integrity**: Check that the SSE/streaming proxy logic doesn't buffer unnecessarily or break the rehydration flow.
           3. **Security**: Look for potential PII leaks or insecure token handling in the vault.
           4. **Reliability**: Ensure the proxy handles upstream errors gracefully.
-          
+
           Keep the summary concise but technical. Post findings as a markdown comment on the PR.
-          
+
           **CRITICAL**: Include the string 'Reviewed SHA: $SHA' at the very end of your comment so I can track which commits have been reviewed."
diff --git a/package-lock.json b/package-lock.json
index 66b2e35..7db1326 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -1,12 +1,12 @@
 {
   "name": "@aictrl/hush",
-  "version": "0.1.6",
+  "version": "0.1.7",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@aictrl/hush",
-      "version": "0.1.6",
+      "version": "0.1.7",
       "license": "Apache-2.0",
       "dependencies": {
         "@modelcontextprotocol/sdk": "^1.27.1",
@@ -4197,9 +4197,9 @@
       }
     },
     "node_modules/xml2js": {
-      "version": "0.4.23",
-      "resolved": "https://registry.npmjs.org/xml2js/-/xml2js-0.4.23.tgz",
-      "integrity": "sha512-ySPiMjM0+pLDftHgXY4By0uswI3SPKLDw/i3UXbnO8M/p28zqexCUoPmQFrYD+/1BzhGJSs2i1ERWKJAtiLrug==",
+      "version": "0.6.2",
+      "resolved": "https://registry.npmjs.org/xml2js/-/xml2js-0.6.2.tgz",
+      "integrity": "sha512-T4rieHaC1EXcES0Kxxj4JWgaUQHDk+qwHcYOCFHfiwKz7tOVPLq7Hjq9dM1WCMhylqMEfP7hMcOIChvotiZegA==",
       "license": "MIT",
       "dependencies": {
         "sax": ">=0.6.0",
diff --git a/package.json b/package.json
index e5a2d73..166353d 100644
--- a/package.json
+++ b/package.json
@@ -52,6 +52,9 @@
     "pino": "^10.3.1",
     "pino-pretty": "^13.1.3"
   },
+  "overrides": {
+    "xml2js": "^0.6.2"
+  },
   "devDependencies": {
     "@types/blessed": "^0.1.27",
     "@types/cors": "^2.8.19",

From 534ecc79d926477fcbda224d5590e177aeb15135 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 15:05:06 +0000
Subject: [PATCH 02/10] fix: use global npm bin path for hush in CI review
 workflow

Running bare `hush` in the repo checkout resolves to the local
package.json bin entry (dist/cli.js) which doesn't exist in CI
since this workflow doesn't build. Use $(npm prefix -g)/bin/hush
to reliably invoke the globally-installed binary.

Also adds a health-check with ::error:: annotation so the job
fails fast with a clear message if the gateway doesn't start.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 1ca0308..7536b76 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -79,12 +79,14 @@ jobs:
         if: steps.check_changes.outputs.skip != 'true'
         run: |
           npm install -g @aictrl/hush@0.1.7
-          PORT=4000 HUSH_HOST=127.0.0.1 hush &
+          HUSH_BIN="$(npm prefix -g)/bin/hush"
+          PORT=4000 HUSH_HOST=127.0.0.1 "$HUSH_BIN" &
           # Wait for gateway to be ready
           for i in $(seq 1 20); do
             curl -sf http://127.0.0.1:4000/health > /dev/null 2>&1 && break
             sleep 0.5
           done
+          curl -sf http://127.0.0.1:4000/health || { echo "::error::Hush gateway failed to start"; exit 1; }
           echo "Hush gateway running on :4000"
 
       - name: Setup OpenCode

From 60b0940ff647179073ddace4a1532d4cb4cebd09 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 16:03:17 +0000
Subject: [PATCH 03/10] fix: add 10m timeout and debug logging to OpenCode
 review step

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 7536b76..bf68912 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -106,12 +106,15 @@ jobs:
 
       - name: Direct OpenCode Review
         if: steps.check_changes.outputs.skip != 'true'
+        timeout-minutes: 10
         env:
           ZHIPU_API_KEY: ${{ secrets.ZHIPUAI_API_KEY }}
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
           echo "Starting review with GLM-5 for SHA $SHA..."
+          echo "opencode.json:"; cat opencode.json
+          echo "Hush health:"; curl -sf http://127.0.0.1:4000/health || echo "gateway unreachable"
 
           $HOME/.opencode/bin/opencode run --model zai-coding-plan/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
 

From b50a99ee59da34d389e4c7c132d5cdd47a862cfd Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 16:15:44 +0000
Subject: [PATCH 04/10] fix: bump OpenCode review timeout to 15m (was timing
 out mid-review)

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index bf68912..a1cb55f 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -106,7 +106,7 @@ jobs:
 
       - name: Direct OpenCode Review
         if: steps.check_changes.outputs.skip != 'true'
-        timeout-minutes: 10
+        timeout-minutes: 15
         env:
           ZHIPU_API_KEY: ${{ secrets.ZHIPUAI_API_KEY }}
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

From 25befab6ee330ef2cc1332a6918fde1074bf3954 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 16:28:32 +0000
Subject: [PATCH 05/10] fix: add build step to review workflow so tests pass
 when OpenCode runs them

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index a1cb55f..2eaa1a7 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -89,6 +89,10 @@ jobs:
           curl -sf http://127.0.0.1:4000/health || { echo "::error::Hush gateway failed to start"; exit 1; }
           echo "Hush gateway running on :4000"
 
+      - name: Build project
+        if: steps.check_changes.outputs.skip != 'true'
+        run: npm ci && npm run build
+
       - name: Setup OpenCode
         if: steps.check_changes.outputs.skip != 'true'
         env:

From f43c082828e068dbf1ca1392a6eaae22161b41fe Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 16:29:45 +0000
Subject: [PATCH 06/10] fix: tell OpenCode reviewer not to run tests, just read
 code and post comment

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 2eaa1a7..71223af 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -122,12 +122,14 @@ jobs:
 
           $HOME/.opencode/bin/opencode run --model zai-coding-plan/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
 
+          **IMPORTANT**: This is a code review only. Do NOT run tests, npm commands, or build commands. Only read source files and git diffs.
+
           Focus areas:
           1. **Redaction Logic**: Ensure PII patterns are robust and handle edge cases in tool outputs (like JSON or CLI tables).
           2. **Streaming Integrity**: Check that the SSE/streaming proxy logic doesn't buffer unnecessarily or break the rehydration flow.
           3. **Security**: Look for potential PII leaks or insecure token handling in the vault.
           4. **Reliability**: Ensure the proxy handles upstream errors gracefully.
 
-          Keep the summary concise but technical. Post findings as a markdown comment on the PR.
+          Keep the summary concise but technical. Post findings as a single markdown comment on the PR using gh pr comment, then stop.
 
           **CRITICAL**: Include the string 'Reviewed SHA: $SHA' at the very end of your comment so I can track which commits have been reviewed."

From eab204510dccbe4d1216ff496865ad46ff72595f Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 17:13:56 +0000
Subject: [PATCH 07/10] fix: use custom provider to route OpenCode through hush
 proxy

OpenCode's built-in providers ignore options.baseURL from opencode.json
(anomalyco/opencode#5674). Define a custom "hush-zhipu" provider with
npm adapter @ai-sdk/openai-compatible so the baseURL is actually
respected and all LLM traffic routes through the hush gateway.

Verified locally: hush logs show requests at /api/coding/paas/v4/chat/completions
with tokenCount > 0 (PII redaction active).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 12 ++++++++----
 examples/team-config/opencode.json    | 11 +++++++++--
 2 files changed, 17 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 71223af..eb5be14 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -103,10 +103,14 @@ jobs:
           curl -fsSL https://opencode.ai/install | bash -s -- --no-modify-path
           echo "$HOME/.opencode/bin" >> $GITHUB_PATH
 
-          # Configure OpenCode to route through hush proxy + hush plugin
+          # Configure OpenCode with custom provider routing through hush proxy.
+          # NOTE: Overriding baseURL on built-in providers (zai-coding-plan) does not
+          # work — OpenCode's bundled @ai-sdk/openai-compatible ignores options.baseURL
+          # (see anomalyco/opencode#5674). Instead, we define a new custom provider
+          # "hush-zhipu" that explicitly sets baseURL via the npm adapter.
           mkdir -p .opencode/plugins
           cp examples/team-config/.opencode/plugins/hush.ts .opencode/plugins/hush.ts
-          printf '%s\n' '{"provider":{"zai-coding-plan":{"options":{"baseURL":"http://127.0.0.1:4000/api/coding/paas/v4"}}},"plugin":[".opencode/plugins/hush.ts"]}' > opencode.json
+          printf '%s\n' '{"provider":{"hush-zhipu":{"npm":"@ai-sdk/openai-compatible","name":"Hush ZhipuAI Proxy","options":{"baseURL":"http://127.0.0.1:4000/api/coding/paas/v4","apiKey":"{env:ZHIPU_API_KEY}"},"models":{"glm-5":{"name":"GLM 5"},"glm-4.7-flash":{"name":"GLM 4.7 Flash"}}}},"plugin":[".opencode/plugins/hush.ts"]}' > opencode.json
 
       - name: Direct OpenCode Review
         if: steps.check_changes.outputs.skip != 'true'
@@ -116,11 +120,11 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
-          echo "Starting review with GLM-5 for SHA $SHA..."
+          echo "Starting review with hush-zhipu/glm-5 for SHA $SHA..."
           echo "opencode.json:"; cat opencode.json
           echo "Hush health:"; curl -sf http://127.0.0.1:4000/health || echo "gateway unreachable"
 
-          $HOME/.opencode/bin/opencode run --model zai-coding-plan/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
+          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
 
           **IMPORTANT**: This is a code review only. Do NOT run tests, npm commands, or build commands. Only read source files and git diffs.
 
diff --git a/examples/team-config/opencode.json b/examples/team-config/opencode.json
index 059a077..27baff8 100644
--- a/examples/team-config/opencode.json
+++ b/examples/team-config/opencode.json
@@ -1,8 +1,15 @@
 {
   "provider": {
-    "zai-coding-plan": {
+    "hush-zhipu": {
+      "npm": "@ai-sdk/openai-compatible",
+      "name": "Hush ZhipuAI Proxy",
       "options": {
-        "baseURL": "http://127.0.0.1:4000/api/coding/paas/v4"
+        "baseURL": "http://127.0.0.1:4000/api/coding/paas/v4",
+        "apiKey": "{env:ZHIPU_API_KEY}"
+      },
+      "models": {
+        "glm-5": { "name": "GLM 5" },
+        "glm-4.7-flash": { "name": "GLM 4.7 Flash" }
       }
     }
   },

From 54675cbad8cc7d4bb7809762389445088f16b980 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 17:32:00 +0000
Subject: [PATCH 08/10] fix: switch review model to glm-4.7-flash for CI speed

GLM-5 takes 3-8 minutes per inference, making it impossible to complete
a multi-file code review within the 15-minute timeout. GLM-4.7-flash
responds in seconds and is sufficient for code review quality.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index eb5be14..6fd4791 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -120,11 +120,13 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
-          echo "Starting review with hush-zhipu/glm-5 for SHA $SHA..."
+          echo "Starting review with hush-zhipu/glm-4.7-flash for SHA $SHA..."
           echo "opencode.json:"; cat opencode.json
           echo "Hush health:"; curl -sf http://127.0.0.1:4000/health || echo "gateway unreachable"
 
-          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-5 "Review the changes in this PR for the Hush Semantic Gateway.
+          # Use glm-4.7-flash for speed — glm-5 takes 3-8 min per inference and
+          # can't complete a multi-file review within the 15-minute timeout.
+          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-4.7-flash "Review the changes in this PR for the Hush Semantic Gateway.
 
           **IMPORTANT**: This is a code review only. Do NOT run tests, npm commands, or build commands. Only read source files and git diffs.
 

From 5215f0de393a6d85378fd2273aa62fe956dc2675 Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 17:50:06 +0000
Subject: [PATCH 09/10] fix: pass PR number explicitly and bump proxy timeout
 to 120s
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

OpenCode wasted 12 minutes trying to discover the PR number via
`gh pr view` (fails in detached HEAD) and `gh pr list --branch`
(flag not available in CI's gh version). Pass $PR_NUMBER directly
in the prompt so it can post the comment immediately.

Also bump the proxy fetch timeout from 30s to 120s — LLM first-token
latency can exceed 30s for large prompts.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 12 +++++++-----
 src/index.ts                          |  2 +-
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 6fd4791..926e278 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -120,13 +120,12 @@ jobs:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
-          echo "Starting review with hush-zhipu/glm-4.7-flash for SHA $SHA..."
+          PR_NUMBER=${{ github.event.pull_request.number }}
+          echo "Starting review with hush-zhipu/glm-4.7-flash for PR #$PR_NUMBER SHA $SHA..."
           echo "opencode.json:"; cat opencode.json
           echo "Hush health:"; curl -sf http://127.0.0.1:4000/health || echo "gateway unreachable"
 
-          # Use glm-4.7-flash for speed — glm-5 takes 3-8 min per inference and
-          # can't complete a multi-file review within the 15-minute timeout.
-          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-4.7-flash "Review the changes in this PR for the Hush Semantic Gateway.
+          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-4.7-flash "Review the changes in PR #$PR_NUMBER for the Hush Semantic Gateway.
 
           **IMPORTANT**: This is a code review only. Do NOT run tests, npm commands, or build commands. Only read source files and git diffs.
 
@@ -136,6 +135,9 @@ jobs:
           3. **Security**: Look for potential PII leaks or insecure token handling in the vault.
           4. **Reliability**: Ensure the proxy handles upstream errors gracefully.
 
-          Keep the summary concise but technical. Post findings as a single markdown comment on the PR using gh pr comment, then stop.
+          Keep the review concise. Post findings as a single markdown comment using:
+            gh pr comment $PR_NUMBER --body \"<your review>\"
+
+          Do NOT try to auto-detect the PR number — use exactly $PR_NUMBER.
 
           **CRITICAL**: Include the string 'Reviewed SHA: $SHA' at the very end of your comment so I can track which commits have been reviewed."
diff --git a/src/index.ts b/src/index.ts
index 0f4cd24..297a6e1 100644
--- a/src/index.ts
+++ b/src/index.ts
@@ -105,7 +105,7 @@ async function proxyRequest(
       method: req.method,
       headers: fetchHeaders,
       body: hasBody ? JSON.stringify(redactedBody) : undefined,
-      signal: AbortSignal.timeout(30000), // 30s timeout
+      signal: AbortSignal.timeout(120000), // 120s timeout (LLM first-token can be slow)
     });
 
     // Handle Upstream Errors (4xx, 5xx)

From 37a0e99f51c3d0f3060e5b8a8ee8fd2e1af4fe3b Mon Sep 17 00:00:00 2001
From: AICtrl Bot <bot@aictrl.dev>
Date: Mon, 2 Mar 2026 18:19:45 +0000
Subject: [PATCH 10/10] fix: switch review model back to glm-5

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .github/workflows/opencode-review.yml | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/.github/workflows/opencode-review.yml b/.github/workflows/opencode-review.yml
index 926e278..d519936 100644
--- a/.github/workflows/opencode-review.yml
+++ b/.github/workflows/opencode-review.yml
@@ -121,11 +121,11 @@ jobs:
         run: |
           SHA=${{ github.event.pull_request.head.sha || github.sha }}
           PR_NUMBER=${{ github.event.pull_request.number }}
-          echo "Starting review with hush-zhipu/glm-4.7-flash for PR #$PR_NUMBER SHA $SHA..."
+          echo "Starting review with hush-zhipu/glm-5 for PR #$PR_NUMBER SHA $SHA..."
           echo "opencode.json:"; cat opencode.json
           echo "Hush health:"; curl -sf http://127.0.0.1:4000/health || echo "gateway unreachable"
 
-          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-4.7-flash "Review the changes in PR #$PR_NUMBER for the Hush Semantic Gateway.
+          $HOME/.opencode/bin/opencode run --model hush-zhipu/glm-5 "Review the changes in PR #$PR_NUMBER for the Hush Semantic Gateway.
 
           **IMPORTANT**: This is a code review only. Do NOT run tests, npm commands, or build commands. Only read source files and git diffs.