bstein/bstein-dev-home
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

portal: add atlasbot profiles

Browse Source
This commit is contained in:
Brad Stein 2026-01-28 13:01:54 -03:00
parent 3ea68a7464
commit 3a27cc9fc1
3 changed files with 149 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

66
backend/atlas_portal/routes/ai.py
View File

@ -19,34 +19,42 @@ def register(app) -> None:
payload = request.get_json(silent=True) or {}
user_message = (payload.get("message") or "").strip()
history = payload.get("history") or []
profile = (payload.get("profile") or payload.get("mode") or "atlas-quick").strip().lower()
if not user_message:
return jsonify({"error": "message required"}), 400
started = time.time()
atlasbot_reply = _atlasbot_answer(user_message)
if atlasbot_reply:
if profile in {"stock", "stock-ai", "stock_ai"}:
reply = _stock_answer(user_message, history)
source = "stock"
else:
mode = "smart" if profile in {"atlas-smart", "smart"} else "quick"
reply = _atlasbot_answer(user_message, mode)
source = f"atlas-{mode}"
if reply:
elapsed_ms = int((time.time() - started) * 1000)
return jsonify({"reply": atlasbot_reply, "latency_ms": elapsed_ms, "source": "atlasbot"})
return jsonify({"reply": reply, "latency_ms": elapsed_ms, "source": source})
elapsed_ms = int((time.time() - started) * 1000)
return jsonify(
{
"reply": "Atlasbot is busy. Please try again in a moment.",
"latency_ms": elapsed_ms,
"source": "atlasbot",
"source": source,
}
)
@app.route("/api/chat/info", methods=["GET"])
@app.route("/api/ai/info", methods=["GET"])
def ai_info() -> Any:
meta = _discover_ai_meta()
profile = (request.args.get("profile") or "atlas-quick").strip().lower()
meta = _discover_ai_meta(profile)
return jsonify(meta)
_start_keep_warm()
def _atlasbot_answer(message: str) -> str:
def _atlasbot_answer(message: str, mode: str) -> str:
endpoint = settings.AI_ATLASBOT_ENDPOINT
if not endpoint:
return ""
@ -55,7 +63,7 @@ def _atlasbot_answer(message: str) -> str:
headers["X-Internal-Token"] = settings.AI_ATLASBOT_TOKEN
try:
with httpx.Client(timeout=settings.AI_ATLASBOT_TIMEOUT_SEC) as client:
resp = client.post(endpoint, json={"prompt": message}, headers=headers)
resp = client.post(endpoint, json={"prompt": message, "mode": mode}, headers=headers)
if resp.status_code != 200:
return ""
data = resp.json()
@ -64,13 +72,55 @@ def _atlasbot_answer(message: str) -> str:
except (httpx.RequestError, ValueError):
return ""
def _discover_ai_meta() -> dict[str, str]:
def _stock_answer(message: str, history: list[dict[str, Any]]) -> str:
body = {
"model": settings.AI_CHAT_MODEL,
"messages": _build_messages(message, history),
"stream": False,
}
try:
with httpx.Client(timeout=settings.AI_CHAT_TIMEOUT_SEC) as client:
resp = client.post(f"{settings.AI_CHAT_API}/api/chat", json=body)
resp.raise_for_status()
data = resp.json()
except (httpx.RequestError, ValueError):
return ""
message_data = data.get("message") if isinstance(data, dict) else None
if isinstance(message_data, dict) and message_data.get("content"):
return str(message_data["content"]).strip()
if isinstance(data, dict) and data.get("response"):
return str(data["response"]).strip()
return ""
def _build_messages(message: str, history: list[dict[str, Any]]) -> list[dict[str, str]]:
messages = [{"role": "system", "content": settings.AI_CHAT_SYSTEM_PROMPT}]
for entry in history:
role = entry.get("role")
content = entry.get("content")
if role in {"user", "assistant"} and isinstance(content, str) and content.strip():
messages.append({"role": role, "content": content})
messages.append({"role": "user", "content": message})
return messages
def _discover_ai_meta(profile: str) -> dict[str, str]:
meta = {
"node": settings.AI_NODE_NAME,
"gpu": settings.AI_GPU_DESC,
"model": settings.AI_CHAT_MODEL,
"endpoint": settings.AI_PUBLIC_ENDPOINT or "/api/chat",
"profile": profile,
}
if profile in {"atlas-smart", "smart"}:
meta["model"] = settings.AI_ATLASBOT_MODEL_SMART or settings.AI_CHAT_MODEL
meta["endpoint"] = "/api/ai/chat"
elif profile in {"atlas-quick", "quick"}:
meta["model"] = settings.AI_ATLASBOT_MODEL_FAST or settings.AI_CHAT_MODEL
meta["endpoint"] = "/api/ai/chat"
elif profile in {"stock", "stock-ai", "stock_ai"}:
meta["model"] = settings.AI_CHAT_MODEL
meta["endpoint"] = "/api/ai/chat"
sa_path = Path("/var/run/secrets/kubernetes.io/serviceaccount")
token_path = sa_path / "token"

2
backend/atlas_portal/settings.py
View File

@ -29,6 +29,8 @@ AI_CHAT_TIMEOUT_SEC = float(os.getenv("AI_CHAT_TIMEOUT_SEC", "20"))
AI_ATLASBOT_ENDPOINT = os.getenv("AI_ATLASBOT_ENDPOINT", "").strip()
AI_ATLASBOT_TOKEN = os.getenv("AI_ATLASBOT_TOKEN", "").strip()
AI_ATLASBOT_TIMEOUT_SEC = float(os.getenv("AI_ATLASBOT_TIMEOUT_SEC", "5"))
AI_ATLASBOT_MODEL_FAST = os.getenv("AI_ATLASBOT_MODEL_FAST", "").strip()
AI_ATLASBOT_MODEL_SMART = os.getenv("AI_ATLASBOT_MODEL_SMART", "").strip()
AI_NODE_NAME = os.getenv("AI_CHAT_NODE_NAME") or os.getenv("AI_NODE_NAME") or "ai-cluster"
AI_GPU_DESC = os.getenv("AI_CHAT_GPU_DESC") or "local GPU (dynamic)"
AI_PUBLIC_ENDPOINT = os.getenv("AI_PUBLIC_CHAT_ENDPOINT", "https://chat.ai.bstein.dev/api/chat")

125
frontend/src/views/AiView.vue
View File

@ -12,16 +12,16 @@
<div class="hero-facts">
<div class="fact">
<span class="label mono">Model</span>
<span class="value mono">{{ meta.model }}</span>
<span class="value mono">{{ current.meta.model }}</span>
</div>
<div class="fact">
<span class="label mono">GPU</span>
<span class="value mono">{{ meta.gpu }}</span>
<span class="value mono">{{ current.meta.gpu }}</span>
</div>
<div class="fact">
<span class="label mono">Endpoint</span>
<button class="endpoint-copy mono" type="button" @click="copyCurl">
{{ meta.endpoint || apiDisplay }}
{{ current.meta.endpoint || apiDisplay }}
<span v-if="copied" class="copied">copied</span>
</button>
</div>
@ -29,8 +29,20 @@
</section>
<section class="card chat-card">
<div class="profile-tabs">
<button
v-for="profile in profiles"
:key="profile.id"
type="button"
class="profile-tab mono"
:class="{ active: activeProfile === profile.id }"
@click="activeProfile = profile.id"
>
{{ profile.label }}
</button>
</div>
<div class="chat-window" ref="chatWindow">
<div v-for="(msg, idx) in messages" :key="idx" :class="['chat-row', msg.role]">
<div v-for="(msg, idx) in current.messages" :key="idx" :class="['chat-row', msg.role]">
<div class="bubble" :class="{ streaming: msg.streaming }">
<div class="role mono">{{ msg.role === 'assistant' ? 'Atlas AI' : 'you' }}</div>
<p class="message">{{ msg.content }}</p>
@ -38,10 +50,10 @@
<div v-else-if="msg.latency_ms" class="meta mono">{{ msg.latency_ms }} ms</div>
</div>
</div>
<div v-if="error" class="chat-row error">
<div v-if="current.error" class="chat-row error">
<div class="bubble">
<div class="role mono">error</div>
<p>{{ error }}</p>
<p>{{ current.error }}</p>
</div>
</div>
</div>
@ -65,32 +77,49 @@
</template>
<script setup>
import { onMounted, onUpdated, ref } from "vue";
import { computed, onMounted, onUpdated, reactive, ref, watch } from "vue";
const API_URL = (import.meta.env.VITE_AI_ENDPOINT || "/api/chat").trim();
const apiUrl = new URL(API_URL, window.location.href);
const apiDisplay = apiUrl.host + apiUrl.pathname;
const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
const meta = ref({
model: "loading...",
gpu: "local GPU (dynamic)",
node: "unknown",
endpoint: apiUrl.toString(),
});
const messages = ref([
{
role: "assistant",
content: "Hi! I'm Atlas AI. How can I help?",
},
]);
const profiles = [
{ id: "atlas-quick", label: "Atlas Quick" },
{ id: "atlas-smart", label: "Atlas Smart" },
{ id: "stock-ai", label: "Stock AI" },
];
const activeProfile = ref("atlas-quick");
const profileState = reactive(
Object.fromEntries(
profiles.map((profile) => [
profile.id,
{
meta: {
model: "loading...",
gpu: "local GPU (dynamic)",
node: "unknown",
endpoint: apiUrl.toString(),
},
messages: [
{
role: "assistant",
content: "Hi! I'm Atlas AI. How can I help?",
},
],
error: "",
},
])
)
);
const current = computed(() => profileState[activeProfile.value]);
const draft = ref("");
const sending = ref(false);
const error = ref("");
const chatWindow = ref(null);
const copied = ref(false);
onMounted(fetchMeta);
onMounted(() => fetchMeta(activeProfile.value));
watch(activeProfile, (profile) => fetchMeta(profile));
onUpdated(() => {
if (chatWindow.value) {
@ -98,16 +127,16 @@ onUpdated(() => {
}
});
async function fetchMeta() {
async function fetchMeta(profile) {
try {
const resp = await fetch("/api/ai/info");
const resp = await fetch(`/api/ai/info?profile=${encodeURIComponent(profile)}`);
if (!resp.ok) return;
const data = await resp.json();
meta.value = {
model: data.model || meta.value.model,
gpu: data.gpu || meta.value.gpu,
node: data.node || meta.value.node,
endpoint: data.endpoint || meta.value.endpoint || apiDisplay,
current.value.meta = {
model: data.model || current.value.meta.model,
gpu: data.gpu || current.value.meta.gpu,
node: data.node || current.value.meta.node,
endpoint: data.endpoint || current.value.meta.endpoint || apiDisplay,
};
} catch {
// swallow
@ -118,20 +147,21 @@ async function sendMessage() {
if (!draft.value.trim() || sending.value) return;
const text = draft.value.trim();
draft.value = "";
error.value = "";
const state = current.value;
state.error = "";
const userEntry = { role: "user", content: text };
messages.value.push(userEntry);
state.messages.push(userEntry);
const assistantEntry = { role: "assistant", content: "", streaming: true };
messages.value.push(assistantEntry);
state.messages.push(assistantEntry);
sending.value = true;
try {
const history = messages.value.filter((m) => !m.streaming).map((m) => ({ role: m.role, content: m.content }));
const history = state.messages.filter((m) => !m.streaming).map((m) => ({ role: m.role, content: m.content }));
const start = performance.now();
const resp = await fetch(API_URL, {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({ message: text, history }),
body: JSON.stringify({ message: text, history, profile: activeProfile.value }),
});
const contentType = resp.headers.get("content-type") || "";
@ -164,7 +194,7 @@ async function sendMessage() {
await typeReveal(assistantEntry, textReply);
}
} catch (err) {
error.value = err.message || "Unexpected error";
state.error = err.message || "Unexpected error";
assistantEntry.content = assistantEntry.content || "(no response)";
assistantEntry.streaming = false;
} finally {
@ -191,7 +221,7 @@ function handleKeydown(e) {
}
async function copyCurl() {
const target = meta.value.endpoint || apiUrl.toString();
const target = current.value.meta.endpoint || apiUrl.toString();
const curl = `curl -X POST ${target} -H 'content-type: application/json' -d '{\"message\":\"hi\"}'`;
try {
await navigator.clipboard.writeText(curl);
@ -264,11 +294,34 @@ async function copyCurl() {
.chat-card {
margin-top: 18px;
display: grid;
grid-template-rows: 1fr auto;
grid-template-rows: auto 1fr auto;
gap: 12px;
min-height: 60vh;
}
.profile-tabs {
display: flex;
flex-wrap: wrap;
gap: 8px;
}
.profile-tab {
border: 1px solid var(--card-border);
background: rgba(255, 255, 255, 0.03);
color: var(--text-muted);
padding: 6px 12px;
border-radius: 999px;
cursor: pointer;
transition: border-color 0.2s ease, color 0.2s ease, background 0.2s ease;
}
.profile-tab.active {
border-color: rgba(0, 229, 197, 0.6);
color: var(--text-primary);
background: rgba(0, 229, 197, 0.12);
box-shadow: var(--glow-soft);
}
.chat-window {
background: rgba(255, 255, 255, 0.02);
border: 1px solid var(--card-border);