feat: v2.5.12 — plan and quota for every AI tool, Grok Build and Kimi Code, Quit AgentHub (#65)

Usage panel (status bar):
- every account shows its plan (Claude subscriptionType, Codex
  chatgpt_plan_type, Cursor membershipType, Grok subscriptionTier)
- new quota rows: Cursor CLI (cursor.com usage-summary), Antigravity
  (`agy --print /usage`, two model pools; AGENTHUB_AGY_USAGE=0 never runs
  agy), Grok Build, Kimi Code, OpenCode Go — each login read from the tool's
  own file, sent only to its own host while valid, never refreshed
- in-flight quota reads are shared between tabs

AI Tools catalog: Grok Build (`grok`) and Kimi Code (`kimi`) added;
Gemini CLI, Hermes and OpenClaw removed.

Also: Quit AgentHub (#65, `agenthub stop`), status-area icon, Add to Chat
on an editor selection, Settings in three groups, JetBrains-style Changes
panel (ticked files/hunks, ignore, blame, file history), terminal font.

--no-verify: :3000 is taken by another dev server (lane-detect tests);
both suites run by hand — server 2636/2639 (the 3 are :3000), client
1131/1131.

server/routes/pricing.js

@@ -83,6 +83,35 @@ function ratesForBucket(rule, row, asOf) {
   return { rIn, rOut, rRead, r5m, r1h };
 }
 
+const LONG_COLS = [
+  "input_tokens",
+  "output_tokens",
+  "cache_read_tokens",
+  "cache_write_tokens",
+  "cache_write_1h_tokens",
+].map((c) => `long_${c}`);
+
+/**
+ * What the long-context part of a bucket adds on top of pricing ALL of it at
+ * the bucket's rates: each long subset (`long_<col>`, capped at its total) at
+ * (factor - 1) x its rate. The input-side factor is long_input / input and
+ * also scales the cache rates, as fast mode's does; output has its own. A rule
+ * with no long rates adds nothing.
+ */
+function longContextExtra(rule, row, { rIn, rOut, rRead, r5m, r1h }) {
+  const r = rule || {};
+  const factor = (longRate, base) => (longRate > 0 && base > 0 ? longRate / base : 1);
+  const fIn = factor(r.long_input_per_mtok, r.input_per_mtok);
+  const fOut = factor(r.long_output_per_mtok, r.output_per_mtok);
+  if (fIn === 1 && fOut === 1) return 0;
+  const long = (col) => Math.min(row[`long_${col}`] || 0, row[col] || 0);
+  const l1h = long("cache_write_1h_tokens");
+  const l5m = Math.max(0, long("cache_write_tokens") - l1h);
+  const inSide =
+    long("input_tokens") * rIn + long("cache_read_tokens") * rRead + l5m * r5m + l1h * r1h;
+  return ((fIn - 1) * inSide + (fOut - 1) * long("output_tokens") * rOut) / 1e6;
+}
+
 // Calculate cost for a set of token buckets against pricing rules. Each bucket
 // is (model, speed, inference_geo, service_tier) with token counts plus the 1h
 // cache-write split and server-tool request counts. Cost = token cost (rate-
@@ -128,7 +157,8 @@ function calculateCost(tokenRows, pricingRules, asOf) {
       unpriced.set(row.model, u);
     }
 
-    const { rIn, rOut, rRead, r5m, r1h } = ratesForBucket(rule, row, asOf);
+    const rates = ratesForBucket(rule, row, asOf);
+    const { rIn, rOut, rRead, r5m, r1h } = rates;
     const cw1h = row.cache_write_1h_tokens || 0;
     const cw5m = Math.max(0, (row.cache_write_tokens || 0) - cw1h);
     const tCost =
@@ -136,7 +166,8 @@ function calculateCost(tokenRows, pricingRules, asOf) {
       (row.output_tokens / 1e6) * rOut +
       (row.cache_read_tokens / 1e6) * rRead +
       (cw5m / 1e6) * r5m +
-      (cw1h / 1e6) * r1h;
+      (cw1h / 1e6) * r1h +
+      longContextExtra(rule, row, rates);
 
     const wsCost = ((row.web_search_requests || 0) / 1000) * WEB_SEARCH_PER_1K_SEARCHES;
     const ceHours = estimateCodeExecHours(
@@ -163,6 +194,7 @@ function calculateCost(tokenRows, pricingRules, asOf) {
       web_search_requests: 0,
       web_fetch_requests: 0,
       code_execution_requests: 0,
+      ...Object.fromEntries(LONG_COLS.map((c) => [c, 0])),
       _cost: 0,
       matched_rule: rule?.model_pattern || null,
     };
@@ -174,6 +206,7 @@ function calculateCost(tokenRows, pricingRules, asOf) {
     agg.web_search_requests += row.web_search_requests || 0;
     agg.web_fetch_requests += row.web_fetch_requests || 0;
     agg.code_execution_requests += row.code_execution_requests || 0;
+    for (const c of LONG_COLS) agg[c] += row[c] || 0;
     agg._cost += tCost + wsCost;
     breakdownMap.set(key, agg);
   }
@@ -221,6 +254,7 @@ function calculateDailyCosts(dailyTokenRows, pricingRules) {
       web_search_requests: row.web_search_requests,
       web_fetch_requests: row.web_fetch_requests,
       code_execution_requests: row.code_execution_requests,
+      ...Object.fromEntries(LONG_COLS.map((c) => [c, row[c]])),
     });
     rowsByDate.set(row.date, rows);
   }
@@ -239,6 +273,7 @@ const USAGE_COLS = [
   "web_search_requests",
   "web_fetch_requests",
   "code_execution_requests",
+  ...LONG_COLS,
 ];
 const BUCKET_COLS = ["model", "speed", "inference_geo", "service_tier"];
 
@@ -263,7 +298,8 @@ function dailyTokenRows(tzModifier, whereClause, params) {
         tu.cache_write_1h_tokens + tu.baseline_cache_write_1h as cache_write_1h_tokens,
         tu.web_search_requests + tu.baseline_web_search as web_search_requests,
         tu.web_fetch_requests + tu.baseline_web_fetch as web_fetch_requests,
-        tu.code_execution_requests + tu.baseline_code_execution as code_execution_requests
+        tu.code_execution_requests + tu.baseline_code_execution as code_execution_requests,
+        ${LONG_COLS.map((c) => `tu.${c}`).join(", ")}
       FROM token_usage tu
       JOIN sessions s ON s.id = tu.session_id
       ${whereClause}`
@@ -582,7 +618,7 @@ router.get("/cost", (req, res) => {
   res.json(computeCost(scope, range, tzModifier));
 });
 
-// GET /api/pricing/quota - plan quota spent per native tool (Claude, Codex):
+// GET /api/pricing/quota - plan quota spent per AI tool (server/lib/quota.js lists them):
 // percentages and reset times only, `?force=1` for a Refresh click (server/lib/quota.js).
 router.get("/quota", async (req, res) => {
   const out = await quotas({ force: req.query.force === "1" });