Phase Timeline¶
| # |
Process |
Start (IST) |
End (IST) |
Duration |
Input |
Output |
Cache Create |
Cache Read |
Tokens |
Cost |
Model |
| 1 |
Requirement scope detection (retried ×1 — exec.crash) (re-dispatched ×2; first attempt 12-09-2026 19:29:07) |
12-09-2026 19:29:16 |
12-09-2026 19:29:40 |
23s |
3 |
1,247 |
31,391 |
29,546 |
62,187 |
$0.2179 |
claude-sonnet-5 |
| 2 |
Initialise run ticket (timing anchor) |
12-09-2026 19:29:40 |
12-09-2026 19:29:42 |
1s |
not available |
not available |
not available |
not available |
not available |
not available |
Script |
| 3 |
Probe environment |
12-09-2026 19:29:42 |
12-09-2026 19:30:01 |
19s |
not available |
not available |
not available |
not available |
not available |
not available |
API Script |
| 4 |
Task Granularity |
12-09-2026 19:30:17 |
12-09-2026 19:31:15 |
57s |
2,548 |
3,853 |
34,065 |
29,157 |
69,623 |
$0.2797 |
claude-opus-5 |
| 5 |
Initialise timing note |
12-09-2026 19:31:15 |
12-09-2026 19:31:15 |
0s |
not available |
not available |
not available |
not available |
not available |
not available |
API Script |
| 6 |
Generate test-spec scenarios and assertions |
12-09-2026 19:31:16 |
12-09-2026 19:31:36 |
19s |
2,545 |
1,033 |
14,041 |
22,055 |
39,674 |
$0.1152 |
claude-sonnet-5 |
| 7 |
Task Granularity — human gate (re-dispatched ×2; first attempt 12-09-2026 19:31:36) |
12-09-2026 19:33:59 |
12-09-2026 19:34:30 |
31s |
n/a |
n/a |
n/a |
n/a |
n/a |
n/a |
n/a |
| 8 |
GAP — account 2 usage_limit cooldown |
12-09-2026 19:34:30 |
17-09-2026 19:48:36 |
120h 14m |
n/a |
n/a |
n/a |
n/a |
n/a — wait time, not compute |
n/a — wait time, not compute |
n/a |
| 9 |
Extract ticket-spec, acceptance criteria & test scenarios (re-dispatched ×2; first attempt 12-09-2026 19:30:01) — cumulative across all 2 attempts: 39,366 tokens, $0.2370 FAILED |
17-09-2026 19:48:36 |
17-09-2026 19:48:36 |
0s |
not available |
not available |
not available |
not available |
not available |
not available |
not recorded |
| — |
ORCHESTRATOR TOTAL (rows 7) |
— |
— |
31s |
0 |
0 |
0 |
0 |
0 |
$0.0000 |
— |
| — |
SHARED TOTAL (rows 1–9) |
— |
— |
2m |
5,096 |
6,133 |
79,497 |
80,758 |
171,484 |
$0.6128 |
— |
| — |
GRAND TOTAL (wall clock) |
12-09-2026 19:29:16 |
17-09-2026 19:48:36 |
120h 19m |
5,096 |
6,133 |
79,497 |
80,758 |
171,484 |
$0.6128 |
— |
ORCHESTRATOR TOTAL: 31s, 0 tokens, $0.0000 (rows 7)
SHARED TOTAL: 2m, 171,484 tokens, $0.6128, 4 row(s) with no token data (rows 1–9)
Total Run Time: 120h 19m (wall clock, 12-09-2026 19:29:16 → 17-09-2026 19:48:36 IST — not a sum of the per-side totals above, since sides overlap)
Active time: 5m | Waiting (session limits, approvals, dead sessions): 120h 14m
Total Tokens Consumed: 171,484 (plus 4 row(s) whose usage was never captured — run execution_tracker.py enrich-audit to backfill. This is a floor, not the total.)
Token breakdown: input 5,096 · output 6,133 · cache-create 79,497 · cache-read 80,758 (47% of the total). Cache-read is context a node re-sends to itself across its own tool-use turns, billed at roughly a tenth of fresh input — a large Tokens number that is mostly cache-read is a long tool loop, not a large workload.
Total Cost: $0.6128 (plus 4 row(s) whose usage was never captured — run execution_tracker.py enrich-audit to backfill. This is a floor, not the total.)
Discarded retry spend: 39,366 tokens, $0.2370 — real, billed spend from attempts that failed or died before completing, on nodes that were later re-dispatched. Not included in Total Tokens/Cost above (those two numbers reflect only each row's final attempt); this is the true total across every attempt combined: 210,850 tokens, $0.8498.