[api-consumption] 📊 GitHub API Consumption Report — 2026-06-20

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-06-20 · Repository: github/gh-aw · Run: #27869705722

⚠️ Data limitation (today): The GitHub REST API consumption field (github_rate_limit_usage.core_consumed) was absent from all 169 downloaded run_summary.json files this run, and token_usage.jsonl files were empty. Today's per-run API quota consumption could not be measured and has not been fabricated — it is recorded as null in the trending history. Run-health metrics below are real. API trend charts are rendered from the real 31-day historical cache; the last day with measured API data is 2026-06-19.

---

Today at a Glance

Metric	Value
🤖 Total Runs	114 (110 ✅ / 4 ❌)
🎯 Success Rate	96.5%
🔗 GitHub REST API Calls	unavailable today (core_consumed field missing from logs — last measured 2026-06-19)
📝 Safe-Output Writes	unavailable today (SafeItemsCount not populated in current log output)
⏱ Avg Duration	531.1s (p95: 967.2s)

---

🔗 GitHub API Calls Trend (90 days)

Over the 31 measured days, daily REST API consumption is volatile, swinging between ~2.4k (2026-06-15) and a dramatic 574k peak on 2026-06-17 that dominates the 7-day rolling average. Excluding that outlier, most days sit in the 7k–80k band, with the most recent measured day (06-19) at a modest 9.8k.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

Among measured days, the heaviest consumers rotate between CI and triage-style workflows (Smoke CI, Issue Monster, PR Sous Chef, AI Moderator). No single workflow holds a permanent lead; quota leadership shifts day to day with event volume rather than a structural hotspot.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Consumption clusters on mid-week development days, with the week containing 2026-06-17 standing out sharply due to that day's 574k spike. Weekend cells are consistently lighter, mirroring lower automation-triggering activity.

---

🍩 Top API Burners (most recent measured day, 2026-06-19)

On the last measured day, consumption was broadly distributed — Smoke CI, Issue Monster, PR Sous Chef and PR Description Updater each took a similar slice rather than one workflow dominating. Concentration risk is low; no single workflow exceeded a few hundred core calls.

---

🔗 GitHub REST API Consumption by Workflow (last measured 24h, 2026-06-19)

The top consumers on the last measured day (Smoke CI ~222, Issue Monster ~221, PR Sous Chef ~203) are all comfortably far from the 15,000/hr core limit. There is no immediate optimisation pressure, though Smoke CI's per-run frequency makes it the natural first target if quota ever tightens.

---

Top 10 Workflows by Run Volume (last 24h)

API-call columns show unavailable because per-run core_consumed was missing from today's logs; run counts and durations are real.

Workflow	REST API Calls	Runs	Avg Duration
PR Sous Chef	unavailable	9	465.8s
PR Code Quality Reviewer	unavailable	7	682.0s
Matt Pocock Skills Reviewer	unavailable	7	530.1s
Design Decision Gate 🏗️	unavailable	7	303.0s
Test Quality Sentinel	unavailable	7	416.3s
Smoke CI	unavailable	5	241.6s
Issue Monster	unavailable	3	302.0s
Contribution Check	unavailable	3	514.3s
PR Description Updater	unavailable	2	399.5s
Auto-Triage Issues	unavailable	2	229.5s

---

Trending Indicators

• 7-day API trend: ↑ +40.8% vs. previous 7 days (skewed by the 2026-06-17 spike; based on measured days only)
• 30-day API trend: → (insufficient data — fewer than 60 measured days in history)
• GitHub REST API call rate: ~97,578 calls/day averaged over the last 7 measured days (hourly limit: 15,000) — note this 7-day mean is inflated by the single 574k outlier on 06-17; the median measured day is far lower

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (31 entries)
• Collection mode: incremental
• Logs start_date used: -1d
• Data points stored: 32 (today appended with github_api_calls: null to avoid fabricating the missing metric)
• Earliest entry: 2026-05-18
• Retention policy: 90 days

🔎 Collection notes

• Logs collected via the agentic-workflows MCP logs tool: 1 initial call (start_date=-1d) + 2 continuation calls (the maximum allowed); each returned a continuation token (timeout-based paging). Collection stopped at the 2-continuation limit with a partial-but-sufficient dataset of 170 run records (169 with run_summary.json).
• Of 169 run summaries, 114 were created on 2026-06-20 and 55 on 2026-06-19.
• Missing fields in this run's log output: github_rate_limit_usage.core_consumed (0/169), populated token_usage.jsonl (0/169), SafeItemsCount > 0 (0/169). These are required for the API and safe-output sections and appear to be a transient gap in the enriched log download — they were present on 2026-06-19.

---

Automatically generated by the api-consumption-report workflow.

Generated by 📊 GitHub API Consumption Report Agent · 229.8 AIC · ⌖ 37.8 AIC · ⊞ 12.1K · ◷

• expires on Jun 23, 2026, 3:42 AM UTC-08:00

[pelikhan]

/q review agentic workflows based on the report and GitHub api rate limit found on workflow runs.

[github-actions[bot]]

🔧 Pay attention, 007! Q is preparing your gadgets for this discussion comment...

[github-actions[bot]]

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

A newer discussion is available at Discussion #40609.