[api-consumption] 📊 GitHub API Consumption Report — 2026-08-23

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-08-23 · Repository: github/gh-aw · Run: #32634768398

⚠️ Data coverage note: This is the first run of this trending pipeline. Log collection is capped at 3 MCP calls per policy (1 initial + 2 continuations), and given the very high current run volume (~100 runs per 5–6 hours), those calls only reached back to 2026-08-22T17:30Z, covering ~17.2 hours of runs rather than a full 24h or 90-day window. All figures below reflect that observed window, not a fixed 24h period. Charts are annotated accordingly and should not be read as full 90-day/30-day history yet — the cache will accumulate real days from here forward.

---

Today at a Glance

Metric	Value
🤖 Total Runs	322 (235 ✅ / 87 ❌)
🎯 Success Rate	72.98%
🔗 GitHub REST API Calls	378,954 (core quota consumed — includes reads, writes, and all GitHub API operations)
📝 Safe-Output Writes	410 (issues + PRs + comments + discussions created by safe-output tools)
⏱ Avg Duration	475.5s (p95: 993.5s)

---

🔗 GitHub API Calls Trend (90 days)

Since the trending cache was just seeded, this chart shows hourly API call volume across the actual observed window (2026-08-22 17:00 UTC → 2026-08-23 10:00 UTC) rather than a true 90-day history. Volume rises sharply through the morning of 2026-08-23, consistent with a burst of PR/issue-review workflow activity; no anomalous spikes beyond the general daytime ramp-up are visible in this short window.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

With only ~17 hours of real history, this shows hourly API consumption for the top 7 workflows rather than a 30-day trend. PR Sous Chef is consistently the heaviest consumer across the window, with several code-review workflows (Impeccable Skills Reviewer, Matt Pocock Skills Reviewer, PR Code Quality Reviewer) clustered close behind — a pattern worth re-checking once 30 real days of data have accumulated.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Only 2 calendar days (2026-08-22, 2026-08-23) are represented, and only the hours within the ~17h observed window are populated — remaining cells are genuinely zero-data (not yet collected), not zero-activity. The populated hours show sustained load rather than a single narrow peak, with the busiest hours falling in the 2026-08-23 morning UTC block.

---

🍩 Top API Burners (24h)

PR Sous Chef alone accounts for 60,973 of the 378,954 total API calls (~16%), with the next five workflows (Impeccable Skills Reviewer, Matt Pocock Skills Reviewer, PR Code Quality Reviewer, Ponytail Reviewer, Design Decision Gate) each consuming roughly 27k calls apiece. Together the top 6 workflows account for the large majority of quota use, indicating moderate concentration risk in PR-review-style workflows rather than a single dominant outlier.

---

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

The top consumer, PR Sous Chef, used 60,973 calls across 59 runs (~1,033/run) — nowhere near the 15,000/hr hourly limit on a per-run basis, but its high run frequency makes it the top aggregate consumer. The five review-style workflows clustered around 27k calls each run far fewer times (16–18 runs), suggesting their per-run cost is roughly 1,500–1,700 calls; consolidating overlapping PR-review workflows (Impeccable/Matt Pocock/PR Code Quality/Ponytail reviewers) could meaningfully cut aggregate quota use if their checks overlap.

---

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

Workflow	REST API Calls	Runs	Avg Duration
PR Sous Chef	60,973	59	418.9s
Impeccable Skills Reviewer	27,532	17	386.0s
Matt Pocock Skills Reviewer	27,441	18	411.0s
PR Code Quality Reviewer	27,163	17	397.2s
Ponytail Reviewer	27,145	17	342.3s
Design Decision Gate 🏗️	27,045	16	405.3s
Issue Monster	18,255	21	344.6s
Avenger	10,298	7	416.4s
PR Description Updater	7,912	6	381.3s
Auto-Triage Issues	6,786	4	310.5s

---

Trending Indicators

• 7-day API trend: → insufficient data (only 2 calendar days recorded so far; need 7+ days of history)
• 30-day API trend: → insufficient data (only 2 calendar days recorded so far; need 30+ days of history)
• GitHub REST API call rate: ~528,600 calls/day extrapolated from the observed 17.2h window (378,954 calls ÷ 17.2h × 24h) — treat as a rough estimate, not a stable 7-day average, given this is the first data point. Hourly limit is 15,000; no single hour in the observed window approached that ceiling in aggregate.

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: no (first run — /tmp/gh-aw/cache-memory/trending/ did not exist)
• Collection mode: backfill
• Logs start_date used: -90d (requested; actual coverage limited to ~17.2h by run volume and the 3-call pagination cap — see note at top of report)
• Data points stored: 2
• Earliest entry: 2026-08-22
• Retention policy: 90 days

---

Automatically generated by the api-consumption-report workflow.

Generated by 📊 GitHub API Consumption Report Agent · agent · 141 AIC · ⌖ 12.7 AIC · ⊞ 13.4K · ◷

• expires on Aug 26, 2026, 3:10 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 #55364.