mirror of
https://github.com/theodore-song/polymarket-analyst.git
synced 2026-07-27 16:07:46 +00:00
Fold agent chat into live API
This commit is contained in:
Theodore Song
@@ -1,168 +0,0 @@
|
||||
const MAX_HISTORY = 12;
|
||||
const MAX_POSITIONS = 10;
|
||||
const MAX_SUGGESTIONS = 8;
|
||||
|
||||
function cleanText(value, max = 1200) {
|
||||
return String(value || "").replace(/\s+/g, " ").trim().slice(0, max);
|
||||
}
|
||||
|
||||
function safeNumber(value, fallback = 0) {
|
||||
const n = Number(value);
|
||||
return Number.isFinite(n) ? n : fallback;
|
||||
}
|
||||
|
||||
function compactPosition(pos) {
|
||||
return {
|
||||
question: cleanText(pos.question, 180),
|
||||
side: cleanText(pos.side, 8),
|
||||
entry_price: safeNumber(pos.entry_price),
|
||||
current_price: safeNumber(pos.current_price),
|
||||
value: safeNumber(pos.value),
|
||||
unrealized_pnl: safeNumber(pos.unrealized_pnl),
|
||||
conviction: safeNumber(pos.conviction),
|
||||
category: cleanText(pos.category, 60),
|
||||
opened_at: cleanText(pos.opened_at, 40),
|
||||
url: cleanText(pos.url, 240),
|
||||
};
|
||||
}
|
||||
|
||||
function compactSuggestion(sug) {
|
||||
return {
|
||||
question: cleanText(sug.question, 180),
|
||||
side: cleanText(sug.side, 8),
|
||||
entry_price: safeNumber(sug.entry_price),
|
||||
conviction: safeNumber(sug.conviction),
|
||||
edge: safeNumber(sug.edge),
|
||||
category: cleanText(sug.category, 60),
|
||||
rationale: cleanText(sug.rationale, 240),
|
||||
url: cleanText(sug.url, 240),
|
||||
};
|
||||
}
|
||||
|
||||
function compactHistory(history) {
|
||||
return (Array.isArray(history) ? history : []).slice(-MAX_HISTORY).map((m) => ({
|
||||
role: m && m.role === "user" ? "user" : "assistant",
|
||||
text: cleanText(m && m.text, 900),
|
||||
})).filter((m) => m.text);
|
||||
}
|
||||
|
||||
function outputText(data) {
|
||||
if (data && typeof data.output_text === "string") return data.output_text.trim();
|
||||
const chunks = [];
|
||||
for (const item of data && Array.isArray(data.output) ? data.output : []) {
|
||||
for (const content of Array.isArray(item.content) ? item.content : []) {
|
||||
if (content.type === "output_text" && content.text) chunks.push(content.text);
|
||||
if (content.type === "text" && content.text) chunks.push(content.text);
|
||||
}
|
||||
}
|
||||
return chunks.join("\n").trim();
|
||||
}
|
||||
|
||||
export default async function handler(req, res) {
|
||||
res.setHeader("Cache-Control", "no-store");
|
||||
if (req.method !== "POST") {
|
||||
res.setHeader("Allow", "POST");
|
||||
return res.status(405).json({ ok: false, error: "Method not allowed" });
|
||||
}
|
||||
|
||||
const apiKey = process.env.OPENAI_API_KEY;
|
||||
if (!apiKey) {
|
||||
return res.status(501).json({
|
||||
ok: false,
|
||||
code: "missing_openai_key",
|
||||
error: "Agent AI chat needs OPENAI_API_KEY in Vercel environment variables.",
|
||||
});
|
||||
}
|
||||
|
||||
const body = req.body || {};
|
||||
const question = cleanText(body.question, 1000);
|
||||
if (!question) return res.status(400).json({ ok: false, error: "Question is required." });
|
||||
|
||||
const agent = body.agent || {};
|
||||
const portfolio = body.portfolio || {};
|
||||
const context = {
|
||||
agent: {
|
||||
id: cleanText(agent.id, 40),
|
||||
name: cleanText(agent.name, 80),
|
||||
kind: cleanText(agent.kind, 40),
|
||||
style: cleanText(agent.style || agent.blurb, 800),
|
||||
voice: cleanText(agent.voice, 200),
|
||||
},
|
||||
portfolio: {
|
||||
equity: safeNumber(portfolio.equity),
|
||||
cash: safeNumber(portfolio.cash),
|
||||
return_pct: safeNumber(portfolio.return_pct),
|
||||
pnl: safeNumber(portfolio.pnl),
|
||||
rank: safeNumber(portfolio.rank),
|
||||
open_positions: safeNumber(portfolio.open_positions),
|
||||
last_decision: cleanText(portfolio.last_decision, 700),
|
||||
recent_actions: (Array.isArray(portfolio.recent_actions) ? portfolio.recent_actions : []).slice(-8).map((x) => cleanText(x, 260)),
|
||||
positions: (Array.isArray(portfolio.positions) ? portfolio.positions : []).slice(0, MAX_POSITIONS).map(compactPosition),
|
||||
snapshots: (Array.isArray(portfolio.snapshots) ? portfolio.snapshots : []).slice(-8).map((s) => ({
|
||||
date: cleanText(s.date || s.timestamp, 40),
|
||||
equity: safeNumber(s.equity),
|
||||
return_pct: safeNumber(s.return_pct),
|
||||
open_positions: safeNumber(s.open_positions),
|
||||
})),
|
||||
},
|
||||
leaderboard: (Array.isArray(body.leaderboard) ? body.leaderboard : []).slice(0, 10).map((r) => ({
|
||||
name: cleanText(r.name, 80),
|
||||
return_pct: safeNumber(r.return_pct),
|
||||
equity: safeNumber(r.equity),
|
||||
rank: safeNumber(r.rank),
|
||||
})),
|
||||
suggestions: (Array.isArray(body.suggestions) ? body.suggestions : []).slice(0, MAX_SUGGESTIONS).map(compactSuggestion),
|
||||
history: compactHistory(body.history),
|
||||
};
|
||||
|
||||
const instructions = [
|
||||
`You are ${context.agent.name || "a Polymarket paper-trading agent"} inside Poly Arena.`,
|
||||
"Speak in first person as the selected agent, like a real chat partner.",
|
||||
"Answer the user's exact question instead of repeating a generic performance summary.",
|
||||
"Use the supplied portfolio, positions, leaderboard, suggestions, and chat history as your facts.",
|
||||
"Be specific about trades, risk, performance, and uncertainty when the data is available.",
|
||||
"Do not claim you can guarantee profits or force agents to make money.",
|
||||
"Do not give personalized financial advice. Keep it framed as paper trading, research, or manual review.",
|
||||
"If asked what you will do next, describe likely decision rules, not a guaranteed action.",
|
||||
"Keep responses concise: 2 to 5 short paragraphs or a tight bullet list.",
|
||||
].join("\n");
|
||||
|
||||
const input = [
|
||||
{
|
||||
role: "user",
|
||||
content: [{
|
||||
type: "input_text",
|
||||
text: `Context JSON:\n${JSON.stringify(context)}\n\nUser message:\n${question}`,
|
||||
}],
|
||||
},
|
||||
];
|
||||
|
||||
try {
|
||||
const response = await fetch("https://api.openai.com/v1/responses", {
|
||||
method: "POST",
|
||||
headers: {
|
||||
"Authorization": `Bearer ${apiKey}`,
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
body: JSON.stringify({
|
||||
model: process.env.OPENAI_MODEL || "gpt-5.6-luna",
|
||||
instructions,
|
||||
input,
|
||||
max_output_tokens: 550,
|
||||
}),
|
||||
});
|
||||
|
||||
const data = await response.json().catch(() => ({}));
|
||||
if (!response.ok) {
|
||||
return res.status(response.status).json({
|
||||
ok: false,
|
||||
error: data && data.error && data.error.message ? data.error.message : "OpenAI chat request failed.",
|
||||
});
|
||||
}
|
||||
|
||||
const text = outputText(data);
|
||||
return res.status(200).json({ ok: true, text: text || "I am here, but I could not form a useful answer from the current context." });
|
||||
} catch (err) {
|
||||
return res.status(500).json({ ok: false, error: err && err.message ? err.message : "Agent chat failed." });
|
||||
}
|
||||
}
|
||||
+161
@@ -245,6 +245,163 @@ function validateIntent(intent) {
|
||||
const VALID_TICKET_STATUSES = new Set(["staged", "reviewed", "placed_manually", "skipped", "cancelled"]);
|
||||
const VALID_FILL_ACTIONS = new Set(["BUY", "SELL"]);
|
||||
const VALID_CAPITAL_ACTIONS = new Set(["DEPOSIT", "WITHDRAW", "BUY_AGENT", "SELL_AGENT"]);
|
||||
const AGENT_CHAT_MAX_HISTORY = 12;
|
||||
const AGENT_CHAT_MAX_POSITIONS = 10;
|
||||
const AGENT_CHAT_MAX_SUGGESTIONS = 8;
|
||||
|
||||
function cleanAgentText(value, max = 1200) {
|
||||
return String(value || "").replace(/\s+/g, " ").trim().slice(0, max);
|
||||
}
|
||||
|
||||
function safeAgentNumber(value, fallback = 0) {
|
||||
const n = Number(value);
|
||||
return Number.isFinite(n) ? n : fallback;
|
||||
}
|
||||
|
||||
function compactAgentPosition(pos) {
|
||||
return {
|
||||
question: cleanAgentText(pos.question, 180),
|
||||
side: cleanAgentText(pos.side, 8),
|
||||
entry_price: safeAgentNumber(pos.entry_price),
|
||||
current_price: safeAgentNumber(pos.current_price),
|
||||
value: safeAgentNumber(pos.value),
|
||||
unrealized_pnl: safeAgentNumber(pos.unrealized_pnl),
|
||||
conviction: safeAgentNumber(pos.conviction),
|
||||
category: cleanAgentText(pos.category, 60),
|
||||
opened_at: cleanAgentText(pos.opened_at, 40),
|
||||
url: cleanAgentText(pos.url, 240),
|
||||
};
|
||||
}
|
||||
|
||||
function compactAgentSuggestion(sug) {
|
||||
return {
|
||||
question: cleanAgentText(sug.question, 180),
|
||||
side: cleanAgentText(sug.side, 8),
|
||||
entry_price: safeAgentNumber(sug.entry_price),
|
||||
conviction: safeAgentNumber(sug.conviction),
|
||||
edge: safeAgentNumber(sug.edge),
|
||||
category: cleanAgentText(sug.category, 60),
|
||||
rationale: cleanAgentText(sug.rationale, 240),
|
||||
url: cleanAgentText(sug.url, 240),
|
||||
};
|
||||
}
|
||||
|
||||
function compactAgentHistory(history) {
|
||||
return (Array.isArray(history) ? history : []).slice(-AGENT_CHAT_MAX_HISTORY).map((m) => ({
|
||||
role: m && m.role === "user" ? "user" : "assistant",
|
||||
text: cleanAgentText(m && m.text, 900),
|
||||
})).filter((m) => m.text);
|
||||
}
|
||||
|
||||
function openAIOutputText(data) {
|
||||
if (data && typeof data.output_text === "string") return data.output_text.trim();
|
||||
const chunks = [];
|
||||
for (const item of data && Array.isArray(data.output) ? data.output : []) {
|
||||
for (const content of Array.isArray(item.content) ? item.content : []) {
|
||||
if (content.type === "output_text" && content.text) chunks.push(content.text);
|
||||
if (content.type === "text" && content.text) chunks.push(content.text);
|
||||
}
|
||||
}
|
||||
return chunks.join("\n").trim();
|
||||
}
|
||||
|
||||
async function handleAgentChat(body, res) {
|
||||
const apiKey = process.env.OPENAI_API_KEY;
|
||||
if (!apiKey) {
|
||||
return res.status(501).json({
|
||||
ok: false,
|
||||
code: "missing_openai_key",
|
||||
error: "Agent AI chat needs OPENAI_API_KEY in Vercel environment variables.",
|
||||
});
|
||||
}
|
||||
|
||||
const question = cleanAgentText(body.question, 1000);
|
||||
if (!question) return res.status(400).json({ ok: false, error: "Question is required." });
|
||||
|
||||
const agent = body.agent || {};
|
||||
const portfolio = body.portfolio || {};
|
||||
const context = {
|
||||
agent: {
|
||||
id: cleanAgentText(agent.id, 40),
|
||||
name: cleanAgentText(agent.name, 80),
|
||||
kind: cleanAgentText(agent.kind, 40),
|
||||
style: cleanAgentText(agent.style || agent.blurb, 800),
|
||||
voice: cleanAgentText(agent.voice, 200),
|
||||
},
|
||||
portfolio: {
|
||||
equity: safeAgentNumber(portfolio.equity),
|
||||
cash: safeAgentNumber(portfolio.cash),
|
||||
return_pct: safeAgentNumber(portfolio.return_pct),
|
||||
pnl: safeAgentNumber(portfolio.pnl),
|
||||
rank: safeAgentNumber(portfolio.rank),
|
||||
open_positions: safeAgentNumber(portfolio.open_positions),
|
||||
last_decision: cleanAgentText(portfolio.last_decision, 700),
|
||||
recent_actions: (Array.isArray(portfolio.recent_actions) ? portfolio.recent_actions : []).slice(-8).map((x) => cleanAgentText(x, 260)),
|
||||
positions: (Array.isArray(portfolio.positions) ? portfolio.positions : []).slice(0, AGENT_CHAT_MAX_POSITIONS).map(compactAgentPosition),
|
||||
snapshots: (Array.isArray(portfolio.snapshots) ? portfolio.snapshots : []).slice(-8).map((s) => ({
|
||||
date: cleanAgentText(s.date || s.timestamp, 40),
|
||||
equity: safeAgentNumber(s.equity),
|
||||
return_pct: safeAgentNumber(s.return_pct),
|
||||
open_positions: safeAgentNumber(s.open_positions),
|
||||
})),
|
||||
},
|
||||
leaderboard: (Array.isArray(body.leaderboard) ? body.leaderboard : []).slice(0, 10).map((r) => ({
|
||||
name: cleanAgentText(r.name, 80),
|
||||
return_pct: safeAgentNumber(r.return_pct),
|
||||
equity: safeAgentNumber(r.equity),
|
||||
rank: safeAgentNumber(r.rank),
|
||||
})),
|
||||
suggestions: (Array.isArray(body.suggestions) ? body.suggestions : []).slice(0, AGENT_CHAT_MAX_SUGGESTIONS).map(compactAgentSuggestion),
|
||||
history: compactAgentHistory(body.history),
|
||||
};
|
||||
|
||||
const instructions = [
|
||||
`You are ${context.agent.name || "a Polymarket paper-trading agent"} inside Poly Arena.`,
|
||||
"Speak in first person as the selected agent, like a real chat partner.",
|
||||
"Answer the user's exact question instead of repeating a generic performance summary.",
|
||||
"Use the supplied portfolio, positions, leaderboard, suggestions, and chat history as your facts.",
|
||||
"Be specific about trades, risk, performance, and uncertainty when the data is available.",
|
||||
"Do not claim you can guarantee profits or force agents to make money.",
|
||||
"Do not give personalized financial advice. Keep it framed as paper trading, research, or manual review.",
|
||||
"If asked what you will do next, describe likely decision rules, not a guaranteed action.",
|
||||
"Keep responses concise: 2 to 5 short paragraphs or a tight bullet list.",
|
||||
].join("\n");
|
||||
|
||||
try {
|
||||
const response = await fetch("https://api.openai.com/v1/responses", {
|
||||
method: "POST",
|
||||
headers: {
|
||||
"Authorization": `Bearer ${apiKey}`,
|
||||
"Content-Type": "application/json",
|
||||
},
|
||||
body: JSON.stringify({
|
||||
model: process.env.OPENAI_MODEL || "gpt-5.6-luna",
|
||||
instructions,
|
||||
input: [{
|
||||
role: "user",
|
||||
content: [{
|
||||
type: "input_text",
|
||||
text: `Context JSON:\n${JSON.stringify(context)}\n\nUser message:\n${question}`,
|
||||
}],
|
||||
}],
|
||||
max_output_tokens: 550,
|
||||
}),
|
||||
});
|
||||
|
||||
const data = await response.json().catch(() => ({}));
|
||||
if (!response.ok) {
|
||||
return res.status(response.status).json({
|
||||
ok: false,
|
||||
error: data && data.error && data.error.message ? data.error.message : "OpenAI chat request failed.",
|
||||
});
|
||||
}
|
||||
|
||||
const text = openAIOutputText(data);
|
||||
return res.status(200).json({ ok: true, text: text || "I am here, but I could not form a useful answer from the current context." });
|
||||
} catch (err) {
|
||||
return res.status(500).json({ ok: false, error: err && err.message ? err.message : "Agent chat failed." });
|
||||
}
|
||||
}
|
||||
|
||||
function marketSearchUrl(question) {
|
||||
return `https://polymarket.com/search?query=${encodeURIComponent(String(question || ""))}`;
|
||||
@@ -366,6 +523,10 @@ export default async function handler(req, res) {
|
||||
if (req.method === "POST") {
|
||||
const body = typeof req.body === "string" ? JSON.parse(req.body || "{}") : (req.body || {});
|
||||
|
||||
if (body.action === "agent_chat") {
|
||||
return handleAgentChat(body, res);
|
||||
}
|
||||
|
||||
if (body.action === "ticket") {
|
||||
const ticket = normalizeTicket(body);
|
||||
const ticketErrors = validateTicket(ticket);
|
||||
|
||||
Reference in New Issue
Block a user