Merge remote-tracking branch 'origin/codex/production-intelligence-terminal' into codex/issue-17-sse-heartbeat

# Conflicts:
#	.env.example
#	README.md
#	crucix.config.mjs
#	test/fetch-utils.test.mjs

.env.example

@@ -10,6 +10,7 @@ STALE_ALERT_COOLDOWN_MINUTES=60
 DASHBOARD_URL=
 TERMINAL_ACTIONS_ENABLED=true
 SWEEP_TOKEN=
+SSE_HEARTBEAT_INTERVAL_MS=25000
 TERMINAL_ACTION_RATE_LIMIT_WINDOW_MS=60000
 TERMINAL_ACTION_RATE_LIMIT_MAX=10
 BRIEF_VERBOSITY=standard

README.md

@@ -134,6 +134,7 @@ STALE_ALERT_COOLDOWN_MINUTES=60
 DASHBOARD_URL=https://intelligence.example.internal
 TERMINAL_ACTIONS_ENABLED=true
 SWEEP_TOKEN=
+SSE_HEARTBEAT_INTERVAL_MS=25000
 TERMINAL_ACTION_RATE_LIMIT_WINDOW_MS=60000
 TERMINAL_ACTION_RATE_LIMIT_MAX=10
 BRIEF_VERBOSITY=standard
@@ -234,6 +235,20 @@ Retention, backup, and privacy expectations:
 - Do not commit `runs/` or `.env`. API credentials stay in `.env`; memory stores derived observations, not secrets.
 - If you expose the dashboard through a reverse proxy, protect Terminal Actions and memory queries behind your normal authentication boundary.
 
+#### Reverse Proxy SSE
+
+The dashboard receives live sweep updates from `GET /events` using Server-Sent Events. The server sends `retry: 10000` reconnect guidance and lightweight heartbeat comments every `SSE_HEARTBEAT_INTERVAL_MS` milliseconds so reverse proxies do not close an otherwise idle stream between 15-minute sweeps.
+
+Recommended proxy settings:
+
+| Proxy | Setting |
+| --- | --- |
+| Pangolin / Traefik-style frontends | Keep response streaming enabled and set idle timeouts above `SSE_HEARTBEAT_INTERVAL_MS`. |
+| Nginx | Disable proxy buffering for `/events`, keep `proxy_read_timeout` above the heartbeat interval, and preserve `Connection: keep-alive`. |
+| Cloudflare-style proxies | Keep the heartbeat below common idle cutoffs; the default 25s is intentionally conservative. |
+
+If you raise the heartbeat interval, keep it shorter than the lowest idle timeout in the proxy chain.
+
 #### Scenario Watchlist
 
 Intelligence Terminal can track operator hypotheses across sweeps with a runtime scenario file at `runs/scenarios.json`. On first run, the server creates three disabled starter examples:

crucix.config.mjs

@@ -29,6 +29,7 @@ export default {
   terminalActionsEnabled: boolEnv('TERMINAL_ACTIONS_ENABLED', !!process.env.SWEEP_TOKEN || process.env.NODE_ENV !== 'production'),
   terminalActionRateLimitWindowMs: intEnv('TERMINAL_ACTION_RATE_LIMIT_WINDOW_MS', 60_000),
   terminalActionRateLimitMax: intEnv('TERMINAL_ACTION_RATE_LIMIT_MAX', 10),
+  sseHeartbeatIntervalMs: intEnv('SSE_HEARTBEAT_INTERVAL_MS', 25000),
 
   llm: {
     provider: process.env.LLM_PROVIDER || null, // anthropic | openai | gemini | codex | openrouter | minimax | mistral | ollama | grok

server.mjs

@@ -370,10 +370,24 @@ app.get('/events', (req, res) => {
     'Cache-Control': 'no-cache',
     'Connection': 'keep-alive',
     'Access-Control-Allow-Origin': '*',
+    'X-Accel-Buffering': 'no',
   });
+  res.write('retry: 10000\n');
   res.write('data: {"type":"connected"}\n\n');
+  const heartbeatMs = Math.max(5000, config.sseHeartbeatIntervalMs || 25000);
+  const heartbeat = setInterval(() => {
+    try {
+      res.write(`: heartbeat ${new Date().toISOString()}\n\n`);
+    } catch {
+      clearInterval(heartbeat);
+      sseClients.delete(res);
+    }
+  }, heartbeatMs);
   sseClients.add(res);
-  req.on('close', () => sseClients.delete(res));
+  req.on('close', () => {
+    clearInterval(heartbeat);
+    sseClients.delete(res);
+  });
 });
 
 function broadcast(data) {

test/fetch-utils.test.mjs

@@ -101,6 +101,17 @@ test('safeFetchText returns text and byte count', async () => {
   }
 });
 
+test('SSE endpoint sends reconnect guidance and clears heartbeat timer', () => {
+  const server = readFileSync(new URL('../server.mjs', import.meta.url), 'utf8');
+  const config = readFileSync(new URL('../crucix.config.mjs', import.meta.url), 'utf8');
+  assert.match(config, /sseHeartbeatIntervalMs/);
+  assert.match(server, /retry: 10000\\n/);
+  assert.match(server, /setInterval\(\(\) =>/);
+  assert.match(server, /: heartbeat/);
+  assert.match(server, /clearInterval\(heartbeat\)/);
+  assert.match(server, /X-Accel-Buffering/);
+});
+
 test('intelligence store defines durable memory and prediction lifecycle tables', () => {
   const store = readFileSync(new URL('../lib/intelligence-store.mjs', import.meta.url), 'utf8');
   assert.match(store, /CREATE TABLE IF NOT EXISTS events/);