[api-consumption] 📊 GitHub API Consumption Report — 2026-07-05

[github-actions[bot]]

📊 GitHub API Consumption Report

Report Date: 2026-07-05 · Repository: github/gh-aw · Run: #28738943602

⚠️ Partial window: The logs(-1d) collection response timed out repeatedly; continuation retries were exhausted. The authoritative on-disk dataset is 210 fully-downloaded runs spanning ~8h (07-05 03:14 → 11:15 UTC) — roughly a third of the intended 24h. Today's figures therefore understate a full day; treat all 24h totals as a lower bound.

---

Today at a Glance

Metric	Value
🤖 Total Runs	210 (175 ✅ / 35 ❌)
🎯 Success Rate	83.3%
🔗 GitHub REST API Calls	132,395 (core quota consumed — includes reads, writes, and all GitHub API operations)
📝 Safe-Output Writes	436 (issues + PRs + comments + discussions created by safe-output tools)
⏱ Avg Duration	637.7s (p95: 1171.0s)

---

🔗 GitHub API Calls Trend (90 days)

Daily REST consumption is volatile, oscillating between near-zero maintenance days and heavy PR-review days. Two clear spikes stand out — ~208k on Jun 09 and a ~572k peak on Jun 17 — both far above the ~55k/day baseline. Today's plotted point (132k) is a partial ~8h window and would likely land near 200k+ if the full day had been collected.

---

🔗 GitHub API Calls by Workflow Trend (30 days)

The same handful of PR-triggered reviewers — Impeccable / Matt Pocock Skills Reviewer, Design Decision Gate, Test Quality Sentinel, PR Code Quality Reviewer — dominate the top of the chart on almost every active day. Their consumption tracks PR volume: quiet on weekends, sharply higher on busy weekday review cycles. No single workflow shows runaway growth; the pattern is broad, PR-driven load.

---

🔗 GitHub REST API Calls Heatmap (90 days)

Consumption concentrates on weekdays, with weekend cells consistently pale — consistent with human-driven PR activity. The mid-June week carries the hottest cells (matching the 208k/572k spikes). Weeks with missing days reflect gaps in the collected history rather than true zero-usage days.

---

🍩 Top API Burners (24h)

Consumption is highly concentrated: the top 3 workflows account for 47% of all REST quota and the top 7 for 94%. All seven are PR-triggered code-review agents that each fire on every pull request. The remaining 55 workflows combined consume only ~6% — meaning any optimisation effort should focus squarely on the review agents.

---

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

The seven leaders each exceed the 15k reference line, but only in 24h aggregate across 23–25 runs — per-run consumption stays modest at ~740–920 calls/run, comfortably below the 15,000/hr hourly cap. The real lever is run count × per-run cost: because these agents run on every PR, trimming redundant GitHub reads per review (e.g. caching PR metadata, batching file fetches) would compound across dozens of daily runs. Individual hourly-limit risk is low.

---

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

Workflow	REST API Calls	Runs	Avg Duration
Impeccable Skills Reviewer	21,539	25	12m 8s
Matt Pocock Skills Reviewer	20,773	24	12m 14s
Design Decision Gate 🏗️	19,321	25	7m 26s
Test Quality Sentinel	18,122	24	8m 10s
PR Code Quality Reviewer	16,995	23	15m 7s
PR Sous Chef	13,735	15	10m 28s
Smoke CI	13,600	15	2m 49s
PR Description Updater	2,558	2	9m 12s
Contribution Check	1,884	2	25m 44s
Daily yamllint Fixer	1,427	1	29m 3s

---

Trending Indicators

• 7-day API trend: ↑ +48.6% vs. previous 7 days (understated — today's point is a partial ~8h window)
• 30-day API trend: → insufficient data (needs 60 daily points; history holds 47)
• GitHub REST API call rate: ~68,563 calls/day over the last 7 days (hourly limit: 15,000; observed per-run cost ~740–920 calls, well within budget)

---

Safe-Output Write Breakdown (24h)

436 safe-output items across 210 runs

Type	Count
create_pull_request_review_comment	189
add_comment	98
submit_pull_request_review	81
create_issue	18
create_discussion	12
push_to_pull_request_branch	10
assign_to_agent	10
create_pull_request	6
add_labels	6
update_pull_request	4
create_check_run	2

Review comments and PR reviews dominate — consistent with the PR-review agents being the top API consumers.

---

📦 Cache Memory Status

• Location: /tmp/gh-aw/cache-memory/trending/api-consumption/history.jsonl
• Cache restored from previous run: yes (46 entries)
• Collection mode: incremental
• Logs start_date used: -1d
• Data points stored: 47
• Earliest entry: 2026-05-18
• Retention policy: 90 days
• Collection note: logs(-1d) returned 1 batch (100 runs); continuation call rejig docs #1 plus its 2 retries all timed out returning the response (data still streamed to disk). Retry budget exhausted per policy. Final dataset = 210 complete runs (~8h, 07-05 03:14→11:15 UTC). Partial — the earlier hours of 07-05 and the 07-04 tail were not fetched.

---

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 · 306.5 AIC · ⌖ 42.5 AIC · ⊞ 11.9K · ◷

• expires on Jul 8, 2026, 3:56 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 #43778.