[api-consumption] 📊 GitHub API Consumption Report — 2026-05-31

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-05-31 · Repository: github/gh-aw · Run: #26711084394

Mid-day snapshot — reflects 151 workflow runs completed as of ~11:30 UTC. Roughly 90 additional runs were still in progress at generation time and are not yet counted.

---

Today at a Glance

Metric	Value
🤖 Total Runs	151 (134 ✅ / 10 ❌ / 7 ⏳ in progress)
🎯 Success Rate	93.1% (of completed runs)
🔗 GitHub REST API Calls	59,440 (core quota consumed — includes reads, writes, and all GitHub API operations)
📝 Safe-Output Writes	107 (22 issues + 23 PRs + 47 comments + 15 discussions created by safe-output tools)
⏱ Avg Duration	393s (p95: 964s)

Activity was healthy today: a 93.1% success rate across 144 completed runs, with 59.4K of GitHub's REST API core quota consumed across 71 distinct workflows. No workflow came close to the 15,000 calls/hour ceiling, and the busiest single consumer (Smoke CI) accounted for 9.3K calls spread over 21 runs.

---

🔗 GitHub API Calls Trend (90 days)

Daily REST API consumption has been volatile over the tracked window, swinging between ~18K and a peak of ~118K on 2026-05-27. After a three-day lull (05-28 → 05-30 hovering near 18–33K), today rebounds to 59.4K — slightly above the 7-day rolling average (~47K). The pattern suggests consumption is driven by bursty PR/CI activity rather than a steady baseline.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

The smoke-test family (Smoke CI, Smoke Claude/Gemini/Copilot) and the PR-review workflows (PR Sous Chef, PR Code Quality Reviewer) are the most consistent top consumers across the window. Spikes in the total line track days with heavy PR review and fuzzing activity, while the per-workflow lines stay comparatively stable — indicating the variation comes from how many runs fire, not from any single workflow becoming more expensive per run.

---

🔗 GitHub REST API Calls Heatmap (90 days)

History currently spans 13 days (2026-05-18 → 2026-05-31), below the 14-day threshold for a calendar heatmap, so this panel falls back to today's top workflows by REST API consumption. Once ≥14 days accumulate, a Mon–Sun calendar heatmap will render here to surface weekly cadence and busiest days.

---

🍩 Top API Burners (24h)

Consumption is moderately concentrated: the top 10 workflows account for ~68% (40.6K) of all REST API calls, with the remaining ~32% spread across 61 other workflows. Smoke CI alone is ~16% of the day's total. No single workflow dominates to the point of concentration risk, but the smoke + PR-review cluster is worth keeping an eye on if daily volume grows.

---

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

The heaviest consumers are Smoke CI (9.3K over 21 runs), Test Quality Sentinel (6.1K), and Matt Pocock Skills Reviewer (4.9K). All are comfortably below the red 15,000 calls/hour reference line — these are daily totals, so per-hour pressure is far lower. Smoke CI's high count reflects run frequency (21 runs) rather than per-run cost (~443 calls/run); batching or caching repository metadata reads across smoke runs would be the most impactful optimisation if quota ever becomes a concern.

---

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

Workflow	REST API Calls	Runs	Avg Duration
Smoke CI	9,297	21	101s
Test Quality Sentinel	6,081	7	279s
Matt Pocock Skills Reviewer	4,943	7	345s
PR Code Quality Reviewer	4,666	7	336s
PR Sous Chef	4,086	10	472s
Design Decision Gate	3,186	7	289s
Auto-Triage Issues	2,541	4	284s
PR Description Updater	2,138	4	295s
Chaos PR Bundle Fuzzer	1,990	3	556s
[aw] Failure Investigator (6h)	1,692	2	2,541s

---

Trending Indicators

• 7-day API trend: → (insufficient data — needs ≥14 days, have 13)
• 30-day API trend: → (insufficient data — needs ≥60 days, have 13)
• GitHub REST API call rate: ~46,583 calls/day averaged over the last 7 days (hourly limit: 15,000; daily headroom is large at this rate)

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (12 entries)
• Collection mode: backfill (only today's day recomputed; days 05-18 → 05-30 already complete in history were retained intact, as the 90-day fetch returned complete summaries only for today)
• Logs start_date used: -90d (with a focused -1d follow-up pass)
• Data points stored: 13
• Earliest entry: 2026-05-18
• Retention policy: 90 days

📊 Full Safe-Output Activity (24h)

All safe-output tool calls observed today (not just the four counted in the headline metric):

Safe-output type	Count
add_comment	47
create_pull_request_review_comment	26
create_pull_request	23
create_issue	22
create_discussion	15
submit_pull_request_review	15
update_issue	11
assign_to_agent	9
upload_asset	9
add_labels	6
update_pull_request	4
create_check_run	3
push_to_pull_request_branch	3
close_issue	1
missing_tool	1
noop	44

The headline Safe-Output Writes (107) counts only created issues, PRs, comments, and discussions per the report definition.

---

Automatically generated by the api-consumption-report workflow.

Generated by 📊 GitHub API Consumption Report Agent · opus48 2.4M · ◷

• expires on Jun 3, 2026, 11:38 AM UTC

[github-actions[bot]]

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

A newer discussion is available at Discussion #36236.