[api-consumption] 📊 GitHub API Consumption Report — 2026-06-25

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-06-25 · Repository: github/gh-aw · Run: #28167093357

⚠️ Window note: Today's collection covers a partial ~12h window (2026-06-24T23:22Z → 2026-06-25T11:27Z). The midday segment of 2026-06-24 was lost when the log continuation call timed out; per the collection policy the workflow proceeded with the partial dataset. Day-over-day comparisons below are skewed by this and by inconsistent historical window sizes.

---

Today at a Glance

Metric	Value
🤖 Total Runs	184 (170 ✅ / 14 ❌)
🎯 Success Rate	92.4%
🔗 GitHub REST API Calls	87,653 (core quota consumed — reads, writes, and all GitHub API operations; all 184 runs reported core_consumed)
📝 Safe-Output Writes	unavailable (safe-output counts not present in this collection window)
⏱ Avg Duration	588s / 9.8m (p95: 1,108s / 18.5m)

---

🔗 GitHub API Calls Trend (90 days)

Daily core-quota consumption is highly volatile across the 90-day record, ranging from a few hundred calls on quiet days to a one-off spike of ~574k on 2026-06-17. Today's 87,653 over a 12h window is among the higher readings, driven by sustained PR-review activity overnight and into the morning. The 7-day rolling average remains elevated but is heavily influenced by that mid-June outlier rather than a steady upward shift.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

The PR-quality cluster — Matt Pocock Skills Reviewer, Test Quality Sentinel, PR Code Quality Reviewer, and Design Decision Gate — consistently dominates per-workflow consumption, with Smoke CI and PR Sous Chef close behind. These same workflows reappear at the top nearly every day, confirming that quota usage tracks PR throughput rather than scheduled/daily agents.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Consumption concentrates on weekdays, mirroring contributor PR activity, with noticeably lighter weekend cells. A small number of cells (the 06-17 spike) stand out as anomalies well above the typical daily band. Most days sit in the low-thousands-to-low-tens-of-thousands range.

---

🍩 Top API Burners (24h)

The top three workflows account for 43.4% of all core quota consumed in the window, and the top seven PR/CI workflows make up the overwhelming majority. Consumption is concentrated but not dangerously so — no single workflow is anywhere near the 15,000/hr ceiling, with the largest (Matt Pocock Skills Reviewer) averaging ~970 quota/run across 14 runs.

---

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

The four PR-review workflows each consumed 12k–13.6k quota over 13–14 runs, making them the clearest optimisation targets — caching repeated get_file/get_diff reads or narrowing fetched file sets would yield the biggest savings. Averaged over the 12h window total consumption is ~7,300/hr, comfortably under the 15,000/hr limit, so there is no immediate throttling risk.

---

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

Workflow	REST API Calls	Runs	Avg Duration
Matt Pocock Skills Reviewer	13,612	14	10.4m
Test Quality Sentinel	12,295	14	8.6m
PR Code Quality Reviewer	12,131	14	11.0m
Design Decision Gate 🏗️	12,006	13	5.7m
Smoke CI	8,403	11	4.5m
PR Description Updater	5,207	5	9.4m
PR Sous Chef	4,982	10	8.6m
Daily Sentrux Report	1,446	1	9.6m
Daily Observability Report for AWF Firewall and MCP Gateway	1,294	1	11.1m
Semantic Function Refactoring	1,291	1	10.6m

---

Trending Indicators

• 7-day API trend: ↑ +293.4% vs. previous 7 days (skewed — the prior 7-day window included several low/partial-data days)
• 30-day API trend: → insufficient consistent data (older entries use mixed window sizes; a clean 30/30 split is not yet available)
• GitHub REST API call rate: ~7,300 calls/hr averaged over today's 12h window (hourly limit: 15,000) — ~49% of ceiling, no throttling risk

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (36 entries)
• Collection mode: incremental
• Logs start_date used: -1d (continuation to fill the 24h window timed out → partial 12h dataset)
• Data points stored: 37
• Earliest entry: 2026-05-18
• Retention policy: 90 days

---

Automatically generated by the api-consumption-report workflow.

Generated by 📊 GitHub API Consumption Report Agent · 249.4 AIC · ⌖ 29.2 AIC · ⊞ 7.1K · ◷

• expires on Jun 28, 2026, 3:50 AM UTC-08:00