From aa61bbe484b99f7a5b90b639663449ffb9dad30d Mon Sep 17 00:00:00 2001
From: Story Crater Bot <19826264+Riotpiaole@users.noreply.github.com>
Date: Sun, 6 Sep 2026 06:08:40 -0700
Subject: [PATCH] fix: add LLM_API_URL and LLM_MODEL to portfolio deployment
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Problem: Portfolio pod had OAuth credentials (AUTHENTIK_CLIENT_ID/SECRET)
but was missing LLM_API_URL and LLM_MODEL environment variables.
Chat API could fetch tokens but had no endpoint to call.

Solution: Inject LLM configuration via k8s deployment env vars:
- LLM_API_URL: https://api.riotpiao.com/v1/chat/completions
- LLM_MODEL: qwen2.5:3b-instruct

Testing:
- Token fetch: ✓ portfolio-agent OAuth works
- DNS: ✓ api.riotpiao.com resolves (CoreDNS rewrite)
- Auth: ⚠️  JWT issuer mismatch (portfolio-agent vs api-gw) — separate issue

Impact: Chat API endpoint can now attempt LLM calls with valid auth
---
 infra/portfolio/base/deployment.yaml | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/infra/portfolio/base/deployment.yaml b/infra/portfolio/base/deployment.yaml
index 9d90270..31e8a0a 100644
--- a/infra/portfolio/base/deployment.yaml
+++ b/infra/portfolio/base/deployment.yaml
@@ -42,6 +42,11 @@ spec:
             secretKeyRef:
               name: portfolio-agent-oidc
               key: TOKEN_URL
+        # LLM API configuration
+        - name: LLM_API_URL
+          value: "https://api.riotpiao.com/v1/chat/completions"
+        - name: LLM_MODEL
+          value: "qwen2.5:3b-instruct"
         ports:
         - name: http
           containerPort: 3000