📊 Daily Copilot Token Consumption Report — 2026-03-24

[github-actions[bot]]

Executive Summary

Over the last 30 days, 89 workflow runs across 44 unique Copilot-powered workflows consumed 85,210,999 tokens (~85.2M). The data covers runs from 2026-02-20 through 2026-03-24, with a −48.8% reduction in tokens compared to the previous reporting period (2026-02-20: ~166.5M tokens, 195 runs). This significant drop reflects a lower run count (−54.4%), consistent with the weekend/holiday schedule pattern.

Key Highlights

• 🏆 Highest token consumer: Daily Syntax Error Quality Check — 11.1M tokens across 2 runs (13.0% of total)
• 🔁 Most active workflow: Issue Monster — 12 runs, 5.7M tokens total
• ⚡ Avg tokens per run: 957,427 (~957K)
• 💬 Total agent turns: 1,667 across all runs
• 📉 Trend: Token usage and run count are both significantly lower vs previous period, indicating a lighter schedule day

---

📈 Token Usage & Run Trends

Token consumption has varied significantly across reporting periods — from a high of 237.8M (2026-02-11, 378 runs) to today's 85.2M (89 runs). The current day reflects a lighter workload consistent with a weekday morning schedule snapshot.

📊 Average Tokens per Workflow Run

The average tokens-per-run metric highlights how efficiently each period used the model. The 2026-02-11 peak had ~629K avg tokens/run, while today shows ~957K — meaning fewer but heavier runs.

---

🏆 Top 10 Workflows by Token Consumption

Rank	Workflow	Total Tokens	Runs	Avg Tokens/Run	Avg Turns/Run	Avg Duration
1	Daily Syntax Error Quality Check	11,057,574	2	5,528,787	103.5	15.3m
2	Contribution Check	5,853,302	6	975,550	29.7	5.2m
3	Functional Pragmatist	5,731,788	1	5,731,788	78.0	17.2m
4	Issue Monster	5,658,229	12	471,519	11.7	5.3m
5	Daily Compiler Quality Check	4,091,631	1	4,091,631	48.0	10.6m
6	Smoke Multi PR	4,087,301	1	4,087,301	89.0	10.5m
7	Glossary Maintainer	3,732,011	1	3,732,011	40.0	12.7m
8	jsweep - JavaScript Unbloater	3,668,933	2	1,834,466	29.0	10.9m
9	Code Simplifier	3,354,572	2	1,677,286	31.5	9.9m
10	CI Cleaner	3,353,591	2	1,676,796	35.5	16.7m

---

💡 Insights & Recommendations

High-Token Workflows

1. Daily Syntax Error Quality Check — 11.1M tokens (13.0% of all consumption)

   • Observation: Averaging 103.5 turns/run and 15.3 min, this is the heaviest single workflow. With only 2 runs in the period, it may be processing a large codebase or running exhaustive checks.
   • Recommendation: Review whether the scope of syntax checking can be narrowed to recently changed files rather than full-repo scans. Consider batching checks or adding early-exit logic.

2. Functional Pragmatist — 5.7M tokens (6.7%) in a single run (78 turns, 17.2 min)

   • Observation: A high-depth single run. This refactoring workflow likely performs many code-analysis iterations.
   • Recommendation: Add early termination when no actionable changes are found, and limit scope to files changed in recent PRs.

3. Smoke Multi PR — 4.1M tokens in a single test run (89 turns, 10.5 min)

   • Observation: A smoke test consuming as much as production workflows suggests it may be overly comprehensive or running against a large test fixture.
   • Recommendation: Reduce the number of PRs created in smoke tests; use lightweight fixtures.

Optimization Opportunities

1. Scope-limited quality checks: Workflows like Daily Syntax Error Quality Check and Daily Compiler Quality Check could target only changed files (via git diff --name-only origin/main) instead of full-repo scans.

   • Affected Workflows: Daily Syntax Error Quality Check, Daily Compiler Quality Check
   • Potential Savings: ~30–50% token reduction on these workflows

2. Parallel batching vs sequential turns: Issue Monster (12 runs, avg 11.7 turns) and Contribution Check (6 runs, avg 29.7 turns) use many turns. Restructuring prompts to batch-process items in fewer turns could reduce overhead.

   • Affected Workflows: Issue Monster, Contribution Check, Auto-Triage Issues
   • Potential Savings: ~10–20% reduction in turns-based overhead

3. Smoke test right-sizing: Smoke tests (Smoke Multi PR, Agent Container Smoke Test) collectively consumed ~5.7M tokens. Consider tiered smoke tests — quick sanity checks vs. full integration tests.

   • Affected Workflows: Smoke Multi PR, Agent Container Smoke Test, Smoke Create Cross-Repo PR
   • Potential Savings: ~2M tokens per reporting period

4. Cache-aware workflows: Workflows like Glossary Maintainer and jsweep may re-read large context each run. Using cache-memory to persist intermediate state could reduce re-reading.

Per-Workflow Detailed Statistics (All 44 Workflows)

Workflow	Total Tokens	Runs	Avg Tokens	Avg Turns	Avg Duration	Success Rate
Daily Syntax Error Quality Check	11,057,574	2	5,528,787	103.5	15.3m	100%
Contribution Check	5,853,302	6	975,550	29.7	5.2m	100%
Functional Pragmatist	5,731,788	1	5,731,788	78.0	17.2m	100%
Issue Monster	5,658,229	12	471,519	11.7	5.3m	100%
Daily Compiler Quality Check	4,091,631	1	4,091,631	48.0	10.6m	100%
Smoke Multi PR	4,087,301	1	4,087,301	89.0	10.5m	100%
Glossary Maintainer	3,732,011	1	3,732,011	40.0	12.7m	100%
jsweep - JavaScript Unbloater	3,668,933	2	1,834,466	29.0	10.9m	100%
Code Simplifier	3,354,572	2	1,677,286	31.5	9.9m	100%
CI Cleaner	3,353,591	2	1,676,796	35.5	16.7m	100%
Daily Community Attribution Updater	3,068,724	2	1,534,362	26.5	17.4m	100%
Dead Code Removal Agent	2,989,364	2	1,494,682	25.0	9.9m	100%
Auto-Triage Issues	2,600,214	7	371,459	8.1	4.3m	100%
Update Astro	2,417,234	1	2,417,234	47.0	9.4m	100%
Daily CLI Performance Agent	2,310,469	3	770,156	15.3	6.8m	100%
Dev	1,998,894	2	999,447	23.0	9.8m	100%
CI Optimization Coach	1,775,906	1	1,775,906	26.0	14.9m	100%
Agent Container Smoke Test	1,661,573	11	151,052	4.2	3.8m	100%
CLI Consistency Checker	1,301,401	1	1,301,401	23.0	8.5m	100%
Copilot PR Conversation NLP Analysis	1,296,894	1	1,296,894	23.0	17.8m	100%
GitHub Remote MCP Authentication Test	1,149,859	2	574,930	16.5	4.2m	100%
Slide Deck Maintainer	1,081,408	1	1,081,408	20.0	7.0m	100%
Daily News	1,072,912	1	1,072,912	21.0	9.8m	100%
Daily Copilot PR Merged Report	1,018,769	1	1,018,769	16.0	8.5m	100%
PR Triage Agent	982,416	3	327,472	7.3	5.8m	100%
Daily MCP Tool Concurrency Analysis	923,429	1	923,429	17.0	7.3m	100%
Daily Regulatory Report Generator	860,840	1	860,840	15.0	6.8m	100%
The Daily Repository Chronicle	826,098	1	826,098	15.0	8.0m	100%
Copilot PR Prompt Pattern Analysis	816,713	1	816,713	17.0	6.8m	100%
Weekly Issue Summary	691,035	1	691,035	15.0	8.2m	100%
Architecture Diagram Generator	569,074	1	569,074	12.0	11.9m	0%
Daily Safe Output Integrator	481,537	1	481,537	9.0	5.0m	100%
Repository Quality Improvement Agent	425,409	1	425,409	8.0	6.6m	100%
Draft PR Cleanup	420,162	1	420,162	11.0	4.1m	100%
Repository Tree Map Generator	407,520	1	407,520	10.0	5.7m	100%
Sub-Issue Closer	387,363	1	387,363	9.0	5.0m	100%
GPL Dependency Cleaner (gpclean)	372,201	1	372,201	9.0	3.8m	100%
Constraint Solving — Problem of the Day	323,662	1	323,662	8.0	5.2m	100%
Issue Triage Agent	235,908	1	235,908	7.0	4.1m	100%
The Great Escapi	155,079	2	77,540	2.0	4.2m	100%
Daily Copilot Token Consumption Report	0	1	0	0.0	0.3m	—
Smoke Create Cross-Repo PR	0	1	0	0.0	2.1m	—
Smoke Update Cross-Repo PR	0	1	0	0.0	2.3m	—
Super Linter Report	0	1	0	0.0	3.1m	—

Historical Comparison

Date	Total Tokens	Runs	Avg Tokens/Run
2026-01-15	94,899,840	178	533,145
2026-01-22	101,969,175	239	426,649
2026-01-28	200,676,032	281	714,149
2026-02-11	237,834,034	378	629,191
2026-02-19	119,326,605	83	1,437,669
2026-02-20	166,507,574	195	854,141
2026-03-24	85,210,999	89	957,427

vs 2026-02-20 (previous report): −48.8% tokens, −54.4% runs

The per-run token average has been trending upward — from 426K (Jan 22) to 957K today — suggesting workflows are becoming more complex or tackling larger tasks per run, even as total run counts fluctuate with schedule patterns.

Methodology & Data Quality Notes

Methodology

• Data Source: Pre-downloaded GitHub Actions workflow logs from last 30 days
• Engine Filter: Copilot engine workflows only
• Memory Storage: memory/token-metrics/history.jsonl in repo memory branch
• Analysis Date: 2026-03-24T11:30 UTC
• Historical Data: 7 reporting days tracked
• Cost Data: Not available for this period (cost tracking appears inconsistent in source data; token counts are primary metric)

Data Quality Notes

• 4 workflows reported 0 tokens (likely non-Copilot steps or in-progress at collection time)
• 1 run (Architecture Diagram Generator) concluded with failure
• Daily Copilot Token Consumption Report (this run) shows 0 tokens as it was still in-progress at log collection time
• 84 of 89 runs (94%) have token usage data

---

References:

• §23486673092 — This report's workflow run
• §23485056190 — Daily Syntax Error Quality Check (top consumer)
• §23481932860 — Functional Pragmatist

AI generated by Daily Copilot Token Consumption Report · history

• expires on Mar 27, 2026, 11:35 AM UTC

[github-actions[bot]]

This discussion was automatically closed because it expired on 2026-03-27T11:35:23.744Z.

Closed by Workflow