[api-consumption] 📊 GitHub API Consumption Report — 2026-07-15

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-07-15 · Repository: github/gh-aw · Run: #29410884688

⚠️ Partial dataset: The logs collection timed out (120s) on the initial call and on the one sanctioned continuation retry. Only ~5 hours of the intended 24h window were captured (runs from 2026-07-15 05:54 → 11:09 UTC, newest-first), covering 202 runs. Counts below reflect that busy morning CI window only — the API total exceeds recent full days because this slice captured a high-volume period, not because consumption tripled. Treat today's figures as a lower bound on run count and non-comparable to full-day history.

---

Today at a Glance

Metric	Value
🤖 Total Runs	202 (147 ✅ / 55 ❌) — partial ~5h window
🎯 Success Rate	72.8%
🔗 GitHub REST API Calls	275,797 (core quota consumed — reads, writes, and all GitHub API operations; partial window)
📝 Safe-Output Writes	Unavailable — only usage artifacts were downloaded; safe-output artifacts were not fetched this run
⏱ Avg Duration	566s (p95: 1178s)

---

🔗 GitHub API Calls Trend (90 days)

Daily REST API consumption has trended upward through July, with several days above 150k and a standout spike near Jun 18 (~575k). Today's 275,797 point sits well above the 7-day rolling average (~110k), but it reflects only a partial, high-traffic morning slice — so it reads as a spike that is really a sampling artifact rather than a genuine surge.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

The heaviest consumers over the last 30 days are consistently the PR-facing review workflows — Smoke CI, the Skills Reviewers (Impeccable / Matt Pocock), Design Decision Gate, and PR Code Quality Reviewer. Smoke CI dominates on high-PR days because it fans out many pull-request-triggered runs, each making dozens of GitHub API calls.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Consumption concentrates on weekdays, tracking developer activity and PR volume; weekends are noticeably lighter. The hottest cells align with mid-June and the current July ramp, while sparse/empty cells mark low-activity days or days outside the retained history.

---

🍩 Top API Burners (24h)

Smoke CI alone burned 26.6% of the window's core quota, and the top 3 workflows together account for ~46% — a meaningful concentration risk. The remaining consumption is spread across the review and triage workflows, with a long tail of low-volume daily agents.

---

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

Smoke CI (73k across 45 runs, ~1.6k/run) leads by a wide margin, followed by the Skills Reviewers and Design Decision Gate at ~2k/run. No single run approached the 15,000/hr core limit — the highest per-run averages (~2.3k for Impeccable Skills Reviewer) leave comfortable headroom. The biggest optimisation lever is Smoke CI's fan-out volume rather than any per-run inefficiency; caching PR metadata or trimming redundant reads there would yield the largest quota savings.

---

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

Workflow	REST API Calls	Runs	Avg Duration
Smoke CI	73,253	45	322s
Impeccable Skills Reviewer	29,886	13	472s
Matt Pocock Skills Reviewer	24,887	14	551s
Design Decision Gate 🏗️	24,439	12	461s
PR Code Quality Reviewer	20,951	13	622s
Tidy	17,149	9	658s
Test Quality Sentinel	17,138	13	466s
PR Sous Chef	11,277	6	530s
Auto-Triage Issues	7,197	4	639s
PR Description Updater	6,690	4	468s

---

Trending Indicators

• 7-day API trend: ↑ +56.2% vs. previous 7 days — inflated by today's partial high-traffic sample; not a reliable signal this run
• 30-day API trend: → (insufficient data — 56 of the 60 days needed for a stable 30-day comparison)
• GitHub REST API call rate: ~109,883 core/day over the last 7 days (hourly limit: 15,000; no single run approached it)

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (55 existing entries)
• Collection mode: incremental
• Logs start_date used: -1d
• Data points stored: 56
• Earliest entry: 2026-05-18
• Retention policy: 90 days

⚠️ Data Collection Notes

• The logs MCP tool timed out after 120s on the initial -1d call and again on the single sanctioned continuation retry (before_run_id=29400265361). Continuation retries were exhausted; the dataset is partial.
• 202 run directories with a parsed run_summary.json were used (of ~493 partially-created directories); the rest lacked a downloaded run_summary.json.
• Only the usage artifact was downloaded (not --artifacts all), so per-run safe_output.jsonl files were absent and SafeItemsCount was 0 across all runs. Safe-output write counts are therefore unavailable for this report and are not implied to be zero.
• API consumption uses github_rate_limit_usage.core_consumed from each run's run_summary.json; all 202 runs reported this field.

---

Automatically generated by the api-consumption-report workflow.

Warning

Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• awmgmcpg

To allow these domains, add them to the network.allowed list in your workflow frontmatter:

network:
  allowed:
    - defaults
    - "awmgmcpg"

See Network Configuration for more information.

Generated by 📊 GitHub API Consumption Report Agent · 232.1 AIC · ⌖ 44.6 AIC · ⊞ 11.9K · ◷

• expires on Jul 18, 2026, 3:32 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 #45977.