2d4debd12a
fix(llm): double the per-client token budget to 100k/min
main
Stéphane Tailland
2026-08-29 07:57:53 +02:00
f9b73fda9f
fix(llm): validate tokens against the classic-models realm, not the stale VPS one
Stéphane Tailland
2026-08-28 11:09:39 +02:00
a26e57b2c5
Let the AI gateway read the bodies it is meant to log
Stéphane Tailland
2026-08-27 16:41:34 +02:00
a43c457976
Follow the AI gateway Kustomization into the llm namespace
Stéphane Tailland
2026-08-27 14:49:06 +02:00
e01b15f14c
Move the AI gateway into the llm namespace and drop the inlining
Stéphane Tailland
2026-08-27 14:47:59 +02:00
b8b1b92226
Reconcile the Flux entrypoints instead of applying them by hand
Stéphane Tailland
2026-08-27 14:21:09 +02:00
05cf2bcce1
Turn on Kong's own AI telemetry instead of improvising it
Stéphane Tailland
2026-08-27 14:12:32 +02:00
11edfa8ed1
Capture request and response bodies on the AI gateway
Stéphane Tailland
2026-08-27 12:48:00 +02:00
9893221da0
Propagate trace context on the AI gateway
Stéphane Tailland
2026-08-27 06:48:31 +02:00
b83a41c0d2
Inline OTEL_TRACES_ENDPOINT so the proxy stack can actually deploy
Stéphane Tailland
2026-08-26 16:40:56 +02:00
5a2addb073
Source the control plane from Vault and use a stable ref name
Stéphane Tailland
2026-08-26 16:30:15 +02:00
502012a5b2
Add metrics, traces and audit logs to the AI gateway
Stéphane Tailland
2026-08-25 07:48:11 +02:00
6de76338a2
Drop TLS_ISSUER_NAME, no longer needed since the shared template sources the Gateway cert from Vault instead of cert-manager
Stéphane Tailland
2026-08-21 09:37:40 +02:00
a0c96eab5b
first commit
Stéphane Tailland
2026-08-20 17:28:05 +02:00
Stéphane Tailland