credit-pack hint: patch budget-exceeded message at BUILD time (runtime rootfs is read-only; boot-time patch crashed startup)

start.sh

@@ -172,42 +172,4 @@ echo "Tinfoil proxy PID: $TINFOIL_PROXY_PID"
 # Give the proxy a moment to verify the enclave attestation
 sleep 3
 
-# Append a co-op-specific hint to LiteLLM's budget-exceeded messages so a
-# member who hits their limit is told they can top up with a credit pack.
-# LiteLLM hardcodes these strings in
-# litellm/proxy/auth/auth_checks.py (not configurable via env/settings), so we
-# patch the running copy at boot. Idempotent (checks a marker first) and
-# re-applied every start, so it survives upstream file moves on version bumps
-# — unlike the old build-time `RUN sed` CORS patch, which broke silently when
-# LiteLLM relocated the file.
-CREDIT_PACK_HINT='. You can add credit packs at https://opencollective.com/inference-cooperative'
-AUTH_CHECKS_PY="$(python3 -c 'import litellm.proxy.auth.auth_checks as m; print(m.__file__)')"
-if [ -n "$AUTH_CHECKS_PY" ] && ! grep -q "opencollective.com/inference-cooperative" "$AUTH_CHECKS_PY" 2>/dev/null; then
-    echo "Patching budget-exceeded messages with credit-pack hint..."
-    python3 - "$AUTH_CHECKS_PY" <<'PYEOF'
-import sys
-p = sys.argv[1]
-s = open(p).read()
-hint = ". You can add credit packs at https://opencollective.com/inference-cooperative"
-before = s
-# Team budget message (the one that fires for members, enforced at team level)
-s = s.replace(
-    'message=f"Budget has been exceeded! Team={team_object.team_id} Current cost: {spend}, Max budget: {team_object.max_budget}"',
-    'message=f"Budget has been exceeded! Team={team_object.team_id} Current cost: {spend}, Max budget: {team_object.max_budget}" + "' + hint.replace('"', '\\"') + '"',
-)
-# Key budget message (a few keys carry individual max_budget)
-s = s.replace(
-    'message=f"Budget has been exceeded! Key={key_descriptor} Current cost: {spend}, Max budget: {valid_token.max_budget}"',
-    'message=f"Budget has been exceeded! Key={key_descriptor} Current cost: {spend}, Max budget: {valid_token.max_budget}" + "' + hint.replace('"', '\\"') + '"',
-)
-if s != before:
-    open(p, "w").write(s)
-    print("patched")
-else:
-    print("no match (upstream message format changed) — leaving unpatched")
-PYEOF
-else
-    echo "Budget-exceeded hint already applied (or auth_checks.py not found); skipping."
-fi
-
 exec litellm --config "$CONFIG_FILE" --port 4000 --host 0.0.0.0