agent-orchestrator-kit 0.11.0 → 0.12.0

package/CHANGELOG.md

@@ -4,6 +4,11 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased]
 
+## [0.12.0] - 2026-09-02
+
+### Fixed
+- **Cursor leftover race, multi-root, 4-decimal estimates, and per-phase clock.** `stop` / `afterAgentResponse` run leftover immediately after a successful jsonl append (fail-open, no stdout); `sessionEnd` leftover stays and stays idempotent. Leftover filters `last.threadId` when it is non-empty and stays time-only when it is `null` or `''`. Hook and collect resolve the consumer in a multi-root window (not the first cwd with `.agents` or `openspec`); leftover walks every candidate that has `openspec/changes`. Written `costUsdEstimated` aggregates are rounded to 4 decimals after sums (`2.3911 + 2.8153 + 1.355 === 6.5614`). Each `phases.<phase>` now stores `startedAt` / `endedAt` / `leadTimeMs` from that phase’s sessions (epoch ms); `durationMs` remains the work-time sum and does not clone `totals.leadTimeMs`.
+
 ## [0.11.0] - 2026-09-02
 
 ### Changed
@@ -266,6 +271,7 @@ All notable changes to this project will be documented in this file.
 ### Added
 - Initial release: 5-role orchestration pipeline, `/opsx:*` commands, IDE sync
 
+[0.12.0]: https://github.com/makshc2/agent-orchestrator-kit/compare/v0.11.0...v0.12.0
 [0.11.0]: https://github.com/makshc2/agent-orchestrator-kit/compare/v0.10.0...v0.11.0
 [0.10.0]: https://github.com/makshc2/agent-orchestrator-kit/compare/v0.9.0...v0.10.0
 [0.9.0]: https://github.com/makshc2/agent-orchestrator-kit/compare/v0.8.0...v0.9.0

package/README.md

@@ -74,7 +74,7 @@ npx agent-orchestrator-kit@latest init --profile generic --ci gitlab --spec-veri
 
 See [Installation](#installation) for profile/CI options.
 
-**🔄 Already have the kit installed? Upgrade to latest (session attribution + Cursor estimate/dedup in v0.11.0; Cursor fallback estimate + first-class `costUsdEstimated` in v0.10.0; UTC timestamps + Amp stamp parse, Amp billed `$`, Cursor API estimate, archive auto-collect in v0.9.0; locked Amp/Cursor client in v0.8.0; `## Metrics` self-report + opt-in `--collect` in v0.7.0 — **BREAKING:** `--no-collect` is gone; change metrics in v0.5.0+, factory phases 1–3 in v0.4.0+, lean pipeline / archive CLI in v0.3.0+, handoff CLI in v0.1.14+, Figma PAT in v0.1.11+):**
+**🔄 Already have the kit installed? Upgrade to latest (Cursor leftover race + multi-root + per-phase clock in v0.12.0; session attribution + Cursor estimate/dedup in v0.11.0; Cursor fallback estimate + first-class `costUsdEstimated` in v0.10.0; UTC timestamps + Amp stamp parse, Amp billed `$`, Cursor API estimate, archive auto-collect in v0.9.0; locked Amp/Cursor client in v0.8.0; `## Metrics` self-report + opt-in `--collect` in v0.7.0 — **BREAKING:** `--no-collect` is gone; change metrics in v0.5.0+, factory phases 1–3 in v0.4.0+, lean pipeline / archive CLI in v0.3.0+, handoff CLI in v0.1.14+, Figma PAT in v0.1.11+):**
 
 ```bash
 npx agent-orchestrator-kit@latest update
@@ -806,14 +806,14 @@ Every change accumulates git-tracked `openspec/changes/<name>/metrics.json` —
 - **Session start** — `handoff --restore` writes a `pending` marker (`startedAt`, expected role).
 - **Session end** — `handoff <name>` closes the pending session: duration, closed role, mapped phase (`explore` / `design` / `spec` / `review` / `apply` / `archive`), runtime (local/cloud), tasks snapshot (`n/m`), and spend from flags → self-report → adapters. Persist collect window is `[pending.startedAt, endedAt]`; a late hook after persist goes to leftover of that session, not the next persist. `--input-tokens` / `--output-tokens` / `--total-tokens` / `--cost-usd` override session totals only and do not wipe `spendByPlatform` / `spendByModel`. Amp billed USD comes from `amp threads usage --details` (`Cost: $N` only; missing line → `costUsd: null`), not from converting credits. Cursor always writes a labeled `costUsdEstimated` when tokens exist: grok-4.6/4.5 use xAI API rates (`costSource: api-estimate`, including `-fast` and the 200k cliff); other models use a versioned fallback of $3/1M input + $15/1M output (or $3.50/1M when only `totalTokens` is present, `costSource: api-estimate-fallback`). That estimate is **not** the Cursor invoice and is never added into billed `costUsd`. Timestamps are ISO-8601 UTC (`…Z`). No restore marker? Pass `--started-at <iso>` or the duration stays honestly `null`.
 - **Archive** — successful `archive <name>` always creates or finalizes `metrics.json`, sets `archivedAt`, appends an Archiver session, collects the locked client (Cursor hook / Amp export+usage / Claude JSONL) in `[pending.startedAt, now]` plus leftover of the previous session, and prints the same human summary as `metrics <name>`. `--collect` still runs all three adapters. Leftover apply `## Metrics` that repeats the previous session is ignored so those tokens are not counted twice.
-- **`sessionEnd` leftover** — `scripts/cursor-spend-collect.cjs` also reads the newest `openspec/changes/archive/*-<name>/metrics.json` when the active change folder is gone, so a late hook after archive still attaches.
-- **Cursor `conversationId`** — restore writes `CURSOR_CONVERSATION_ID` into `pending.threadId`; Cursor collect skips rows whose `conversationId` does not match when a filter id is present.
+- **`sessionEnd` leftover** — `scripts/cursor-spend-collect.cjs` also reads the newest `openspec/changes/archive/*-<name>/metrics.json` when the active change folder is gone, so a late hook after archive still attaches. `stop` / `afterAgentResponse` run the same leftover after a successful jsonl append. When `last.threadId` is non-empty, leftover keeps only rows whose `conversationId` matches that id (empty/`null` threadId stays time-only). In a multi-root window leftover walks every candidate that has `openspec/changes`, reading that root’s jsonl. Aggregates write `costUsdEstimated` with 4 decimal places. Each `phases.<phase>` stores `startedAt` / `endedAt` / `leadTimeMs` from that phase’s sessions; `durationMs` stays the work-time sum and does not clone `totals.leadTimeMs`.
+- **Cursor `conversationId`** — restore writes `CURSOR_CONVERSATION_ID` into `pending.threadId`; Cursor collect and sessionEnd leftover skip rows whose `conversationId` does not match when a filter id is present (`last.threadId` on leftover).
 - **Canonical role** — `session.role` stores the first known token (`Explorer`, `Architect`, `Spec Reviewer`, `Implementer`, `Archiver`, `Design Intake`); Closed role in `handoff.md` MAY keep a sentence after `—`.
 - **Platform** — `--platform` → `## Metrics: platform` → `AOK_PLATFORM` → **pending client from `--restore`** → host env (Amp / Cursor / Claude Code) → collected sources (`cursor|claude|amp` only). Invalid `--platform` fails before persist/move.
 - **Locked client** — `--restore` records `pending.platform` and Amp `pending.threadId` before phase work. Persist follows that client’s flow even if persist runs in another shell (no `AMP_*` / `CURSOR_*`). Amp: `amp threads export` plus `amp threads usage --details` (`AOK_AMP_BIN`) and local `threads/*.json`. Export supplies `model` / tokens / `agentMode`; usage supplies billed `$`. If Amp runs tools over a pipe (`/dev/null`), thread id comes from `amp threads list`, not stale `session.json` `lastThreadId`. Cursor: spend hook file. Claude: `~/.claude/projects`. `--collect` still runs all three adapters.
-- **Cursor spend hook (optional)** — Cursor never writes token usage to disk, so the kit can install `scripts/cursor-spend-hook.cjs` plus `.cursor/hooks.json` entries (`stop` / `subagentStop` / `afterAgentResponse`) in `init` / `update` / `sync` / `mcp-setup`: the hook appends each turn's tokens to gitignored `.agents/spend/cursor-usage.jsonl`. Persist auto-reads that file when the locked client is Cursor. Persist and restore do not self-heal the hook. `sessionEnd` still runs `scripts/cursor-spend-collect.cjs`. Restart Cursor once after the first install. `status` shows a `Spend capture` section. Claude JSONL remains a fallback. Amp web/CLI spend is taken from `amp threads export` (tokens, model, `agentMode`) and `amp threads usage` (billed USD).
+- **Cursor spend hook (optional)** — Cursor never writes token usage to disk, so the kit can install `scripts/cursor-spend-hook.cjs` plus `.cursor/hooks.json` entries (`stop` / `subagentStop` / `afterAgentResponse`) in `init` / `update` / `sync` / `mcp-setup`: the hook appends each turn's tokens to gitignored `.agents/spend/cursor-usage.jsonl`. After a successful `stop` / `afterAgentResponse` append the hook runs leftover (fail-open, no stdout). Hook and collect resolve the consumer in a multi-root window (not the first cwd with `.agents` or `openspec`). Persist auto-reads that file when the locked client is Cursor. Persist and restore do not self-heal the hook. `sessionEnd` still runs `scripts/cursor-spend-collect.cjs`. Restart Cursor once after the first install. `status` shows a `Spend capture` section. Claude JSONL remains a fallback. Amp web/CLI spend is taken from `amp threads export` (tokens, model, `agentMode`) and `amp threads usage` (billed USD).
 
-Aggregates are recomputed on every write: per-phase totals (`durationMs`, tokens, `costUsd`, `sessions`, `roles`, `models`) plus overall `totals` (`sessions`, `cloudSessions`, `durationMs` = sum of session work time, `leadTimeMs` = wall clock from first session start to last session end), `spend` (USD only), and separate **by platform** / **by model** tables. Numbers are null-honest: a metric nobody reported stays `null`, never a fake `0`. No single total that adds Amp credits to USD.
+Aggregates are recomputed on every write: per-phase totals (`startedAt`, `endedAt`, `leadTimeMs` from that phase’s sessions, `durationMs` = sum of session work time — not `totals.leadTimeMs` and not `endedAt − startedAt`, tokens, `costUsd`, `costUsdEstimated` to 4 decimals, `sessions`, `roles`, `models`) plus overall `totals` (`sessions`, `cloudSessions`, `durationMs` = sum of session work time, `leadTimeMs` = wall clock from first session start to last session end), `spend` (USD only), and separate **by platform** / **by model** tables. Numbers are null-honest: a metric nobody reported stays `null`, never a fake `0`. No single total that adds Amp credits to USD.
 
 Fill `## Metrics` in `handoff.md` **before** persist (unknown is fine; do not invent `0`):
 
@@ -1006,6 +1006,11 @@ Phase bounds and non-goals: [`openspec/specs/agentic-factory-roadmap/spec.md`](o
 
 ## Changelog
 
+### 0.12.0
+- Cursor leftover after `stop` / `afterAgentResponse` (fail-open, idempotent `sessionEnd`); leftover filters `last.threadId` when it is set
+- Multi-root hook/collect resolve the consumer window; leftover walks every candidate with `openspec/changes`
+- `costUsdEstimated` aggregates write 4 decimals; `phases.<phase>` stores `startedAt` / `endedAt` / `leadTimeMs` (work-time `durationMs` is not `totals.leadTimeMs`)
+
 ### 0.11.0
 - Session metrics attribution: persist collect is `[pending.startedAt, endedAt]`; a late hook stays leftover of that session; archive is `[pending.startedAt, now]` plus leftover of the previous session
 - Source product id wins for `session.model`; `--model` / `## Metrics` / `AOK_MODEL` apply only when sources have none

package/bin/agent-orchestrator.js

@@ -1615,6 +1615,11 @@ const METRICS_VERSION = 1;
 const METRICS_SPEND_KEYS = ['inputTokens', 'outputTokens', 'totalTokens', 'costUsd', 'costUsdEstimated'];
 const LEFTOVER_GRACE_MS = 120000;
 
+function roundUsd4(x) {
+  if (x == null) return null;
+  return Math.round(Number(x) * 10000) / 10000;
+}
+
 function metricsFilePath(projectDir, changeName) {
   return join(projectDir, 'openspec', 'changes', changeName, 'metrics.json');
 }
@@ -1902,7 +1907,7 @@ function sourceEstimatedUsd(sources) {
   for (const src of sources || []) {
     sum = addNullable(sum, numOrNull(src.costUsdEstimated));
   }
-  return sum;
+  return roundUsd4(sum);
 }
 
 function ampThreadTotals(threads) {
@@ -2202,6 +2207,12 @@ function recomputeSpendMaps(metrics) {
     }
     addModelRow(session.model, session.platform, sessionNums);
   }
+  for (const key of Object.keys(byPlatform)) {
+    byPlatform[key].costUsdEstimated = roundUsd4(byPlatform[key].costUsdEstimated);
+  }
+  for (const row of byModel.values()) {
+    row.costUsdEstimated = roundUsd4(row.costUsdEstimated);
+  }
   metrics.spendByPlatform = byPlatform;
   metrics.spendByModel = [...byModel.values()];
 }
@@ -2214,15 +2225,39 @@ function recomputeMetricsAggregates(metrics) {
   let firstStart = null;
   let lastEnd = null;
   for (const session of metrics.sessions) {
+    session.costUsdEstimated = roundUsd4(session.costUsdEstimated);
     totals.sessions += 1;
     if (session.runtime === 'cloud') totals.cloudSessions += 1;
     totals.durationMs = addNullable(totals.durationMs, numOrNull(session.durationMs));
-    if (session.startedAt) firstStart = firstStart == null ? session.startedAt : earlierTimestamp(firstStart, session.startedAt);
-    if (session.endedAt) lastEnd = lastEnd == null ? session.endedAt : laterTimestamp(lastEnd, session.endedAt);
+    if (Number.isFinite(parseFlexibleIso(session.startedAt))) {
+      firstStart = firstStart == null ? session.startedAt : earlierTimestamp(firstStart, session.startedAt);
+    }
+    if (Number.isFinite(parseFlexibleIso(session.endedAt))) {
+      lastEnd = lastEnd == null ? session.endedAt : laterTimestamp(lastEnd, session.endedAt);
+    }
     const key = session.phase || 'other';
-    const phase = phases[key] || { sessions: 0, durationMs: null, ...emptySpendTotals(), agents: [], models: [] };
+    const phase = phases[key] || {
+      sessions: 0,
+      durationMs: null,
+      startedAt: null,
+      endedAt: null,
+      leadTimeMs: null,
+      ...emptySpendTotals(),
+      agents: [],
+      models: [],
+    };
     phase.sessions += 1;
     phase.durationMs = addNullable(phase.durationMs, numOrNull(session.durationMs));
+    if (Number.isFinite(parseFlexibleIso(session.startedAt))) {
+      phase.startedAt = phase.startedAt == null
+        ? session.startedAt
+        : earlierTimestamp(phase.startedAt, session.startedAt);
+    }
+    if (Number.isFinite(parseFlexibleIso(session.endedAt))) {
+      phase.endedAt = phase.endedAt == null
+        ? session.endedAt
+        : laterTimestamp(phase.endedAt, session.endedAt);
+    }
     for (const spendKey of METRICS_SPEND_KEYS) {
       let value = null;
       if ((session.sources || []).length > 0) {
@@ -2247,6 +2282,15 @@ function recomputeMetricsAggregates(metrics) {
   if (firstStart && lastEnd) {
     totals.leadTimeMs = Math.max(0, parseFlexibleIso(lastEnd) - parseFlexibleIso(firstStart));
   }
+  for (const phase of Object.values(phases)) {
+    const startMs = parseFlexibleIso(phase.startedAt);
+    const endMs = parseFlexibleIso(phase.endedAt);
+    phase.leadTimeMs = Number.isFinite(startMs) && Number.isFinite(endMs)
+      ? Math.max(0, endMs - startMs)
+      : null;
+    phase.costUsdEstimated = roundUsd4(phase.costUsdEstimated);
+  }
+  spend.costUsdEstimated = roundUsd4(spend.costUsdEstimated);
   metrics.phases = phases;
   metrics.totals = totals;
   metrics.spend = spend;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agent-orchestrator-kit",
-  "version": "0.11.0",
+  "version": "0.12.0",
   "description": "Universal AI agent orchestration kit for Cursor, Claude Code, and Amp Code — spec-driven OpenSpec pipeline, conductor subagents, durable session handoff, factory gates and MCP setup, cloud-agent handoff, and optional local Figma PAT setup",
   "keywords": [
     "ai-agent",

package/templates/scripts/cursor-spend-collect.cjs

@@ -5,7 +5,7 @@
 'use strict';
 
 const { existsSync, readdirSync, readFileSync, writeFileSync, statSync } = require('fs');
-const { join } = require('path');
+const { join, resolve } = require('path');
 
 function numOrNull(value) {
   if (value == null || value === '') return null;
@@ -18,6 +18,17 @@ function addNullable(a, b) {
   return (a ?? 0) + (b ?? 0);
 }
 
+function roundUsd4(x) {
+  if (x == null) return null;
+  return Math.round(Number(x) * 10000) / 10000;
+}
+
+function timestampMs(value) {
+  if (value == null || value === '') return NaN;
+  const ms = Date.parse(value);
+  return Number.isFinite(ms) ? ms : NaN;
+}
+
 const GROK_46 = {
   inputPerM: 2,
   cachedPerM: 0.5,
@@ -85,14 +96,161 @@ function describeCursorCostEstimate(args) {
   };
 }
 
+function uniqueExistingAbsPaths(payload) {
+  const seen = new Set();
+  const out = [];
+  const add = (value) => {
+    if (value == null || value === '') return;
+    let abs;
+    try {
+      abs = resolve(String(value));
+    } catch {
+      return;
+    }
+    if (seen.has(abs)) return;
+    try {
+      if (!existsSync(abs)) return;
+    } catch {
+      return;
+    }
+    seen.add(abs);
+    out.push(abs);
+  };
+  add(process.cwd());
+  const roots = payload && Array.isArray(payload.workspace_roots) ? payload.workspace_roots : [];
+  for (const root of roots) add(root);
+  return out;
+}
+
+function leftoverCandidateRoots(payload) {
+  return uniqueExistingAbsPaths(payload).filter((root) => existsSync(join(root, 'openspec', 'changes')));
+}
+
+function readJsonSafe(filePath) {
+  try {
+    return JSON.parse(readFileSync(filePath, 'utf-8'));
+  } catch {
+    return null;
+  }
+}
+
+function metricsHasConversationThread(metrics, conversationId) {
+  if (!metrics || typeof metrics !== 'object' || !conversationId) return false;
+  const pendingId = metrics.pending && metrics.pending.threadId != null && metrics.pending.threadId !== ''
+    ? String(metrics.pending.threadId).trim()
+    : '';
+  if (pendingId && pendingId === conversationId) return true;
+  const sessions = Array.isArray(metrics.sessions) ? metrics.sessions : [];
+  const last = sessions[sessions.length - 1];
+  const lastId = last && last.threadId != null && last.threadId !== ''
+    ? String(last.threadId).trim()
+    : '';
+  return Boolean(lastId && lastId === conversationId);
+}
+
+function activeChangeNames(changesDir) {
+  const names = [];
+  let entries;
+  try {
+    entries = readdirSync(changesDir);
+  } catch {
+    return names;
+  }
+  for (const name of entries) {
+    if (name === 'archive') continue;
+    const full = join(changesDir, name);
+    try {
+      if (statSync(full).isDirectory()) names.push(name);
+    } catch {
+      continue;
+    }
+  }
+  return names;
+}
+
+function rootHasMatchingThread(root, conversationId) {
+  const changesDir = join(root, 'openspec', 'changes');
+  if (!existsSync(changesDir)) return false;
+  for (const name of activeChangeNames(changesDir)) {
+    if (metricsHasConversationThread(readJsonSafe(join(changesDir, name, 'metrics.json')), conversationId)) {
+      return true;
+    }
+  }
+  const archiveDir = join(changesDir, 'archive');
+  if (!existsSync(archiveDir)) return false;
+  const seen = new Set();
+  let entries;
+  try {
+    entries = readdirSync(archiveDir);
+  } catch {
+    return false;
+  }
+  for (const entry of entries) {
+    const changeName = archivedChangeName(entry);
+    if (!changeName || seen.has(changeName)) continue;
+    seen.add(changeName);
+    const dir = newestArchiveDirForName(archiveDir, changeName);
+    if (!dir) continue;
+    if (metricsHasConversationThread(readJsonSafe(join(dir, 'metrics.json')), conversationId)) return true;
+  }
+  return false;
+}
+
+function rootHasActiveChange(root) {
+  return activeChangeNames(join(root, 'openspec', 'changes')).length > 0;
+}
+
+function jsonlHasConversationId(root, conversationId) {
+  if (!conversationId) return false;
+  const filePath = join(root, '.agents', 'spend', 'cursor-usage.jsonl');
+  if (!existsSync(filePath)) return false;
+  let text;
+  try {
+    text = readFileSync(filePath, 'utf-8');
+  } catch {
+    return false;
+  }
+  for (const line of text.split('\n')) {
+    if (!line.trim()) continue;
+    let row;
+    try {
+      row = JSON.parse(line);
+    } catch {
+      continue;
+    }
+    const id = row && row.conversationId != null && row.conversationId !== ''
+      ? String(row.conversationId).trim()
+      : '';
+    if (id && id === conversationId) return true;
+  }
+  return false;
+}
+
 function resolveBaseDir(payload) {
-  const cwd = process.cwd();
-  if (existsSync(join(cwd, 'openspec', 'changes'))) return cwd;
-  const roots = Array.isArray(payload.workspace_roots) ? payload.workspace_roots : [];
-  for (const root of roots) {
-    if (root && existsSync(join(String(root), 'openspec', 'changes'))) return String(root);
+  const candidates = uniqueExistingAbsPaths(payload && typeof payload === 'object' ? payload : {});
+  const conversationId = String((payload && payload.conversation_id) || '').trim();
+  if (conversationId) {
+    for (const root of candidates) {
+      if (rootHasMatchingThread(root, conversationId)) return root;
+    }
+  }
+  for (const root of candidates) {
+    if (rootHasActiveChange(root)) return root;
+  }
+  if (conversationId) {
+    for (const root of candidates) {
+      if (jsonlHasConversationId(root, conversationId)) return root;
+    }
+  }
+  const cwd = resolve(process.cwd());
+  if (
+    candidates.includes(cwd)
+    && (existsSync(join(cwd, 'openspec', 'changes')) || existsSync(join(cwd, '.agents')))
+  ) {
+    return cwd;
   }
-  return cwd;
+  if (!candidates.length) return cwd;
+  return [...candidates].sort()[0];
 }
 
 const CURSOR_LEFTOVER_GRACE_MS = 120000;
@@ -255,7 +413,7 @@ function sourceTotals(sources) {
     if (src.costUsd != null) costUsd = addNullable(costUsd, numOrNull(src.costUsd));
     costUsdEstimated = addNullable(costUsdEstimated, numOrNull(src.costUsdEstimated));
   }
-  return { inputTokens, outputTokens, totalTokens, costUsd, costUsdEstimated };
+  return { inputTokens, outputTokens, totalTokens, costUsd, costUsdEstimated: roundUsd4(costUsdEstimated) };
 }
 
 function looksOverridden(session) {
@@ -298,13 +456,22 @@ function recompute(metrics) {
     totals.sessions += 1;
     if (session.runtime === 'cloud') totals.cloudSessions += 1;
     totals.durationMs = addNullable(totals.durationMs, numOrNull(session.durationMs));
-    if (session.startedAt && (firstStart == null || session.startedAt < firstStart)) firstStart = session.startedAt;
-    if (session.endedAt && (lastEnd == null || session.endedAt > lastEnd)) lastEnd = session.endedAt;
+    const sessionStartMs = timestampMs(session.startedAt);
+    if (Number.isFinite(sessionStartMs) && (firstStart == null || sessionStartMs < timestampMs(firstStart))) {
+      firstStart = session.startedAt;
+    }
+    const sessionEndMs = timestampMs(session.endedAt);
+    if (Number.isFinite(sessionEndMs) && (lastEnd == null || sessionEndMs > timestampMs(lastEnd))) {
+      lastEnd = session.endedAt;
+    }
 
     const key = session.phase || 'other';
     const phase = phases[key] || {
       sessions: 0,
       durationMs: null,
+      startedAt: null,
+      endedAt: null,
+      leadTimeMs: null,
       inputTokens: null,
       outputTokens: null,
       totalTokens: null,
@@ -315,6 +482,12 @@ function recompute(metrics) {
     };
     phase.sessions += 1;
     phase.durationMs = addNullable(phase.durationMs, numOrNull(session.durationMs));
+    if (Number.isFinite(sessionStartMs) && (phase.startedAt == null || sessionStartMs < timestampMs(phase.startedAt))) {
+      phase.startedAt = session.startedAt;
+    }
+    if (Number.isFinite(sessionEndMs) && (phase.endedAt == null || sessionEndMs > timestampMs(phase.endedAt))) {
+      phase.endedAt = session.endedAt;
+    }
     for (const spendKey of ['inputTokens', 'outputTokens', 'totalTokens', 'costUsd', 'costUsdEstimated']) {
       const fromSession = numOrNull(session[spendKey]);
       let value = fromSession;
@@ -375,6 +548,21 @@ function recompute(metrics) {
   if (firstStart && lastEnd) {
     totals.leadTimeMs = Math.max(0, Date.parse(lastEnd) - Date.parse(firstStart));
   }
+  for (const phase of Object.values(phases)) {
+    const startMs = timestampMs(phase.startedAt);
+    const endMs = timestampMs(phase.endedAt);
+    phase.leadTimeMs = Number.isFinite(startMs) && Number.isFinite(endMs)
+      ? Math.max(0, endMs - startMs)
+      : null;
+    phase.costUsdEstimated = roundUsd4(phase.costUsdEstimated);
+  }
+  spend.costUsdEstimated = roundUsd4(spend.costUsdEstimated);
+  for (const bucket of Object.values(byPlatform)) {
+    bucket.costUsdEstimated = roundUsd4(bucket.costUsdEstimated);
+  }
+  for (const row of byModel.values()) {
+    row.costUsdEstimated = roundUsd4(row.costUsdEstimated);
+  }
   metrics.phases = phases;
   metrics.totals = totals;
   metrics.spend = spend;
@@ -382,12 +570,19 @@ function recompute(metrics) {
   metrics.spendByModel = [...byModel.values()];
 }
 
-function incomingCursorSources(cwd, existing, fingerprints, windowStart, windowEnd, byId, exclusiveEnd) {
+function incomingCursorSources(cwd, existing, fingerprints, windowStart, windowEnd, byId, exclusiveEnd, filterConversationId) {
   const startMs = windowStart ? Date.parse(windowStart) : NaN;
   const endMs = windowEnd ? Date.parse(windowEnd) : NaN;
+  const filterId = String(filterConversationId || '').trim();
   const bestById = new Map();
   for (const [id, row] of byId) {
     if (existing.has(id)) continue;
+    if (filterId) {
+      const rowConversationId = row.conversationId == null || row.conversationId === ''
+        ? ''
+        : String(row.conversationId).trim();
+      if (rowConversationId !== filterId) continue;
+    }
     const atMs = Date.parse(row.at);
     if (Number.isFinite(startMs) && Number.isFinite(atMs) && atMs < startMs) continue;
     if (Number.isFinite(endMs) && Number.isFinite(atMs)) {
@@ -527,6 +722,7 @@ function backfillMetricsFile(cwd, filePath) {
       leftoverEnd,
       byId,
       leftoverEndExclusive(metrics),
+      last.threadId,
     )
     : [];
   if (incoming.length) {
@@ -602,14 +798,7 @@ function newestArchiveDirForName(archiveDir, changeName) {
   return best;
 }
 
-function main(raw) {
-  let payload = {};
-  try {
-    payload = raw ? JSON.parse(raw) : {};
-  } catch {
-    payload = {};
-  }
-  const cwd = resolveBaseDir(payload && typeof payload === 'object' ? payload : {});
+function backfillRoot(cwd) {
   const changesDir = join(cwd, 'openspec', 'changes');
   if (!existsSync(changesDir)) return;
   const activeNames = new Set();
@@ -645,21 +834,48 @@ function main(raw) {
   }
 }
 
-if (process.stdin.isTTY) {
+function parsePayload(raw) {
+  if (raw && typeof raw === 'object') return raw;
   try {
-    main('');
-  } catch {}
-  process.exit(0);
+    return raw ? JSON.parse(raw) : {};
+  } catch {
+    return {};
+  }
 }
 
-let input = '';
-process.stdin.on('data', (chunk) => {
-  input += chunk;
-});
-process.stdin.on('end', () => {
-  try {
-    main(input);
-  } catch {}
-  process.exit(0);
-});
-process.stdin.on('error', () => process.exit(0));
+function main(raw) {
+  const payload = parsePayload(raw);
+  const candidates = leftoverCandidateRoots(payload && typeof payload === 'object' ? payload : {});
+  for (const cwd of candidates) {
+    try {
+      backfillRoot(cwd);
+    } catch {}
+  }
+}
+
+if (require.main === module) {
+  if (process.stdin.isTTY) {
+    try {
+      main('');
+    } catch {}
+    process.exit(0);
+  }
+
+  let input = '';
+  process.stdin.on('data', (chunk) => {
+    input += chunk;
+  });
+  process.stdin.on('end', () => {
+    try {
+      main(input);
+    } catch {}
+    process.exit(0);
+  });
+  process.stdin.on('error', () => process.exit(0));
+}
+
+module.exports = {
+  main,
+  backfillLeftover: main,
+  resolveBaseDir,
+};

package/templates/scripts/cursor-spend-hook.cjs

@@ -5,7 +5,7 @@
 // must never block the agent loop, so every failure path exits 0 with no output.
 'use strict';
 
-const { existsSync, mkdirSync, appendFileSync } = require('fs');
+const { mkdirSync, appendFileSync } = require('fs');
 const { join } = require('path');
 
 function numOrNull(value) {
@@ -14,14 +14,24 @@ function numOrNull(value) {
   return Number.isFinite(n) ? n : null;
 }
 
+function loadCollect() {
+  return require('./cursor-spend-collect.cjs');
+}
+
 function resolveBaseDir(payload) {
-  const cwd = process.cwd();
-  if (existsSync(join(cwd, '.agents'))) return cwd;
-  const roots = Array.isArray(payload.workspace_roots) ? payload.workspace_roots : [];
-  for (const root of roots) {
-    if (root && existsSync(join(String(root), '.agents'))) return String(root);
-  }
-  return cwd;
+  try {
+    const collect = loadCollect();
+    if (typeof collect.resolveBaseDir === 'function') return collect.resolveBaseDir(payload);
+  } catch {}
+  return process.cwd();
+}
+
+function runLeftover(payload) {
+  try {
+    const collect = loadCollect();
+    const fn = collect.backfillLeftover || collect.main;
+    if (typeof fn === 'function') fn(payload);
+  } catch {}
 }
 
 function main(raw) {
@@ -65,6 +75,11 @@ function main(raw) {
   const spendDir = join(resolveBaseDir(payload), '.agents', 'spend');
   mkdirSync(spendDir, { recursive: true });
   appendFileSync(join(spendDir, 'cursor-usage.jsonl'), `${JSON.stringify(record)}\n`);
+
+  const event = payload.hook_event_name ? String(payload.hook_event_name) : '';
+  if (event === 'stop' || event === 'afterAgentResponse') {
+    runLeftover(payload);
+  }
 }
 
 let input = '';