[api-consumption] 📊 GitHub API Consumption Report — 2026-06-24

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-06-24 · Repository: github/gh-aw · Run: #28095530528

i️ Window note: This run captured 00:00–11:29 UTC (report executes mid-day), so today's totals are a partial-day snapshot and will trend higher by end of day. All 142 runs reported core_consumed, so the API figures below are complete for the captured window.

---

Today at a Glance

Metric	Value
🤖 Total Runs	142 (124 ✅ / 18 ❌)
🎯 Success Rate	87.3%
🔗 GitHub REST API Calls	28,896 (core quota consumed — includes reads, writes, and all GitHub API operations; 82 distinct workflows active)
📝 Safe-Output Writes	Data unavailable this run (no safe_output.jsonl present in downloaded run logs)
⏱ Avg Duration	629.3s (p95: 1290.0s)

---

🔗 GitHub API Calls Trend (90 days)

Daily core-quota consumption remains highly volatile, swinging between ~800 and ~574k on a log scale. The single largest spike was 574,243 on Jun 17 (391 runs), an order of magnitude above the typical 10k–80k band. Today's 28,896 sits comfortably mid-range; the 7-day rolling average (red) has climbed back toward ~100k after the late-May/early-June plateau, driven almost entirely by that Jun 17 outlier.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

Over the last 30 days, PR Sous Chef and the Matt Pocock Skills Reviewer are the most consistent high-consumers, repeatedly appearing in the top tier on a symlog scale. Smoke-test workflows (Smoke Pi, Smoke CI, Smoke Antigravity) spike sporadically rather than steadily, reflecting their on-demand nature. No new workflow has emerged as a sustained heavy consumer beyond the established leaders.

---

🔗 GitHub REST API Calls Heatmap (90 days)

The weekly heatmap shows consumption concentrated on weekdays, with weekends consistently lighter — consistent with development-driven workflow triggers. The standout cell is mid-June (Jun 17), whose intensity dwarfs every other day. Today's Wednesday cell is moderate and partial (collection ended at 11:29 UTC).

---

🍩 Top API Burners (24h)

Consumption is well-distributed rather than concentrated: the top 8 workflows account for ~52% of quota, and the remaining 74 workflows ("Other") make up 47.8%. The single largest burner, PR Sous Chef (9.1%), is closely followed by Matt Pocock Skills Reviewer (8.8%). With no workflow exceeding ~9% of the daily total, there is no meaningful concentration risk today.

---

🔗 GitHub REST API Consumption by Workflow (last 24h)

The top consumers are PR Sous Chef (2,622 across 10 runs) and Matt Pocock Skills Reviewer (2,543 across 6 runs). Crucially, no workflow is anywhere near the 15,000/hour rate limit — the heaviest per-run footprint (~260 calls/run) leaves a wide margin. The smoke-test family collectively consumes a notable share; batching or caching their setup API calls is the most obvious optimisation opportunity if quota pressure ever grows.

---

Top 10 Workflows by REST API Consumption (last 24h)

Workflow	REST API Calls	Runs	Avg Duration
PR Sous Chef	2,622	10	8.5m
Matt Pocock Skills Reviewer	2,543	6	11.1m
AI Moderator	1,842	4	18.1m
Design Decision Gate 🏗️	1,820	6	5.8m
Smoke Pi	1,582	2	6.4m
Smoke Antigravity	1,577	2	5.3m
Agent Container Smoke Test	1,557	2	5.0m
Smoke CI	1,548	6	4.2m
PR Code Quality Reviewer	1,440	6	12.2m
Smoke Claude	1,222	2	13.1m

---

Trending Indicators

• 7-day API trend: ↑ +64.0% vs. previous 7 days (noisy — both windows are heavily skewed by the Jun 17 outlier)
• 30-day API trend: → insufficient data (36 of 60 required data points stored)
• GitHub REST API call rate: ~9,800/day median over the last 7 days (mean 90,900/day is inflated by the 574k Jun 17 spike) — well within the 15,000/hour limit at the per-run level

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (35 entries → 36 after merge)
• Collection mode: incremental
• Logs start_date used: -1d (plus 2 continuation pages to complete the 24h window)
• Data points stored: 36
• Earliest entry: 2026-05-18
• Retention policy: 90 days

⚠️ Data Caveats

• Partial day: collection window was 00:00–11:29 UTC; end-of-day totals will be higher.
• Safe-output writes unavailable: downloaded run logs contained no safe_output.jsonl files this run, so issue/PR/comment/discussion write counts could not be computed.
• Day-over-day comparison skew: the prior entry (Jun 23) was a partial 49-run capture (continuation timed out), so it under-reports relative to today's 142-run window.
• Trend volatility: a single 574,243-quota day (Jun 17) dominates 7-day rolling and mean statistics; medians are more representative.

---

Automatically generated by the api-consumption-report workflow.

Generated by 📊 GitHub API Consumption Report Agent · 262.2 AIC · ⌖ 37 AIC · ⊞ 7.4K · ◷

• expires on Jun 27, 2026, 3:52 AM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by GitHub API Consumption Report Agent.

A newer discussion is available at Discussion #41418.