[api-consumption] 📊 GitHub API Consumption Report — 2026-07-10

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-07-10 · Repository: github/gh-aw · Run: #29089932924

⏱ Partial window: today's figures cover 2026-07-10 00:00–11:30 UTC (~11.5h of the day). The prior day 2026-07-09 is a gap — only 4 late-night runs were retrievable, insufficient for a representative daily entry, so no 07-09 point was recorded.

---

Today at a Glance

Metric	Value
🤖 Total Runs	152 (124 ✅ / 28 ❌)
🎯 Success Rate	81.6%
🔗 GitHub REST API Calls	69,905 (core quota consumed — includes reads, writes, and all GitHub API operations)
📝 Safe-Output Writes	257 (issues + PRs + comments + discussions created by safe-output tools)
⏱ Avg Duration	564.0s (p95: 1129.0s)

---

🔗 GitHub API Calls Trend (90 days)

Daily REST API consumption is highly bursty — the 90-day mean sits near 60,935 calls/day, but individual days swing from under 5,000 to a single 574,243-call spike on 2026-06-17 (flagged as the only >2σ anomaly). July has trended sharply upward, with 07-04 (182,812), 07-05 (132,395) and 07-08 (157,821) all clearing 130k. Today's partial 69,905 already exceeds the 90-day average with half the day still remaining.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

The PR-review family — Impeccable Skills Reviewer, Matt Pocock Skills Reviewer, PR Code Quality Reviewer and Test Quality Sentinel — are the consistent top consumers over the last 30 days, tracking PR activity closely. Smoke CI stays elevated on high-run days, and consumption for all leaders scales with run volume rather than per-run cost, pointing to fan-out (frequency) as the dominant driver.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Consumption concentrates on weekdays, mirroring PR and CI cadence, with visibly lighter weekends. The most recent weeks (late June into July) are the hottest bands, and the Jun-17 outlier stands out as an isolated single-day burst rather than a sustained pattern.

---

🍩 Top API Burners (24h)

A small set of workflows dominates: Smoke CI alone accounts for ~20% of the day's REST API calls, and the top five (Smoke CI plus the four reviewers) together consume over 60% of the total. This concentration means optimising just those five workflows would move the overall footprint the most.

---

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

Smoke CI (13,786 across 14 runs, ~985/run) is the clear leader, followed by four reviewers each in the 8,000–8,800 range. No single workflow-run approached the 15,000/hr core rate limit — consumption is spread across many short runs rather than concentrated in one — so the near-term risk is aggregate throttling on burst days, not per-run exhaustion. Batching read calls in the reviewers and trimming redundant status polls in Smoke CI are the highest-leverage optimisations.

---

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

Workflow	REST API Calls	Runs	Avg Duration
Smoke CI	13,786	14	259.6s
Impeccable Skills Reviewer	8,821	9	380.0s
PR Code Quality Reviewer	8,285	9	512.7s
Test Quality Sentinel	8,110	9	444.2s
Matt Pocock Skills Reviewer	8,071	9	494.9s
Design Decision Gate 🏗️	5,933	9	451.8s
jsweep - JavaScript Unbloater	4,945	1	1155.0s
PR Sous Chef	3,986	10	572.5s
PR Description Updater	2,531	3	447.3s
Dependabot Burner	1,552	1	531.0s

---

Trending Indicators

• 7-day API trend: ↑ +92.9% vs. previous 7 days (driven by the 07-04/07-05/07-08 high-volume days)
• 30-day API trend: → insufficient data (need 60 days of history; 51 stored)
• GitHub REST API call rate: ~92,621 calls/day averaged over the last 7 days (hourly limit: 15,000)

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (50 entries)
• Collection mode: incremental
• Logs start_date used: -2d
• Data points stored: 51
• Earliest entry: 2026-05-18
• Retention policy: 90 days

⚠️ Collection Notes

• The logs(-2d) MCP call timed out during response formatting after 120s (a formatting timeout, not a paging failure). 156 run directories with complete metrics were collected to disk and used as the authoritative dataset (152 for 2026-07-10, 4 for 2026-07-09). No continuation was attempted per guidance for formatting timeouts.
• Today's window is partial (~11.5h); daily totals will be higher once the full day completes.
• 2026-07-09 remains a gap (only 4 unrepresentative late-night runs available).

---

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 · 235.4 AIC · ⌖ 42.3 AIC · ⊞ 11.9K · ◷

• expires on Jul 13, 2026, 3:54 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 #44940.