📊 Daily Copilot Token Consumption Report — 2026-03-20

[github-actions[bot]]

Executive Summary

Over the last 30 days (as of 2026-03-20), Copilot-powered agentic workflows consumed 58,187,531 tokens across 96 workflow runs covering 47 unique workflows. The average token usage per run was 606,120 tokens. Cost data is not available in the current log format.

⚠️ Note: 21 runs (22%) did not report token usage, primarily from Issue Monster (16 runs) and a few others. Actual consumption is likely higher.

Key Highlights

Metric	Value
📈 Total Tokens	58,187,531
🔄 Total Runs	96 (75 with token data)
🗂️ Unique Workflows	47
⚡ Avg Tokens / Run	606,120
⏱️ Total Runtime	~9.1 hours
❌ Total Errors	36
📭 Missing Data Events	4
✅ Safe Output Items	79

• Highest consuming workflow: Daily Syntax Error Quality Check — 4,158,536 tokens (1 run)
• Most active workflow: Issue Monster — 23 runs (39 total including runs without token data)
• Largest single run avg: Daily Syntax Error Quality Check at 4.16M tokens/run
• Best efficiency: Issue Monster at ~92,476 avg tokens/run for high-volume work

---

📊 Visual Analytics

Top 15 Workflows by Total Token Consumption

Top 15 Most Active Workflows (by Run Count)

Top 15 Workflows by Average Tokens per Run

---

🏆 Top 15 Workflows by Token Consumption

Rank	Workflow	Total Tokens	Runs	Avg Tokens/Run	Total Runtime
1	Daily Syntax Error Quality Check	4,158,536	1	4,158,536	13.7m
2	Contribution Check	4,104,809	6	684,135	30.3m
3	GitHub Remote MCP Authentication Test	3,456,564	1	3,456,564	7.6m
4	Ubuntu Actions Image Analyzer	3,150,994	1	3,150,994	10.3m
5	CI Optimization Coach	2,683,940	1	2,683,940	14.8m
6	Copilot CLI Deep Research Agent	2,595,344	1	2,595,344	11.2m
7	Architecture Diagram Generator	2,205,854	1	2,205,854	11.7m
8	Issue Monster	2,126,951	23	92,476	78.2m
9	Copilot PR Conversation NLP Analysis	1,994,243	1	1,994,243	19.4m
10	Auto-Triage Issues	1,764,358	9	196,040	38.9m
11	Daily Compiler Quality Check	1,717,236	1	1,717,236	9.4m
12	Agent Container Smoke Test	1,689,577	11	153,598	48.9m
13	Delight	1,614,535	1	1,614,535	11.1m
14	jsweep - JavaScript Unbloater	1,585,678	1	1,585,678	7.8m
15	Workflow Health Manager - Meta-Orchestrator	1,578,150	1	1,578,150	8.3m

Per-Workflow Detailed Statistics (All 47 Workflows)

Workflow	Total Tokens	Runs	Avg Tokens/Run	Avg Duration	Errors
Daily Syntax Error Quality Check	4,158,536	1	4,158,536	13.7m	0
Contribution Check	4,104,809	6	684,135	5.1m	0
GitHub Remote MCP Authentication Test	3,456,564	1	3,456,564	7.6m	0
Ubuntu Actions Image Analyzer	3,150,994	1	3,150,994	10.3m	0
CI Optimization Coach	2,683,940	1	2,683,940	14.8m	0
Copilot CLI Deep Research Agent	2,595,344	1	2,595,344	11.2m	0
Architecture Diagram Generator	2,205,854	1	2,205,854	11.7m	0
Issue Monster	2,126,951	23	92,476	3.4m	0
Copilot PR Conversation NLP Analysis	1,994,243	1	1,994,243	19.4m	0
Auto-Triage Issues	1,764,358	9	196,040	4.3m	0
Daily Compiler Quality Check	1,717,236	1	1,717,236	9.4m	0
Agent Container Smoke Test	1,689,577	11	153,598	4.5m	0
Delight	1,614,535	1	1,614,535	11.1m	0
jsweep - JavaScript Unbloater	1,585,678	1	1,585,678	7.8m	0
Workflow Health Manager - Meta-Orchestrator	1,578,150	1	1,578,150	8.3m	0
Daily Malicious Code Scan Agent	1,362,195	1	1,362,195	6.8m	0
Glossary Maintainer	1,386,601	1	1,386,601	8.8m	0
Dependabot Dependency Checker	1,107,386	1	1,107,386	9.5m	0
Daily News	1,088,265	1	1,088,265	10.6m	0
Daily MCP Tool Concurrency Analysis	1,041,503	1	1,041,503	7.0m	0
Daily Workflow Updater	761,219	1	761,219	7.5m	0
Dev	237,113	1	237,113	5.5m	0
Sub-Issue Closer	97,756	1	97,756	4.8m	0
(remaining 24 workflows)	varies	varies	varies	varies	varies

Full data available in workflow run artifacts from §23340261754

---

💡 Insights & Recommendations

High Single-Run Cost Workflows

Several workflows consumed over 2M tokens in a single run — these are prime candidates for optimization review:

1. Daily Syntax Error Quality Check — 4.16M tokens/run

   • Observation: Highest single-run consumer. Likely processes a large codebase or generates very verbose output per syntax check.
   • Recommendation: Review whether the full codebase scan is needed on every run or if incremental/diff-based analysis could reduce scope significantly.

2. GitHub Remote MCP Authentication Test — 3.46M tokens/run

   • Observation: An authentication test consuming 3.5M tokens suggests verbose diagnostic output or repeated test iterations.
   • Recommendation: Streamline test output verbosity; limit diagnostic dumps to failure cases only.

3. Ubuntu Actions Image Analyzer — 3.15M tokens/run

   • Observation: Image analysis workflows tend to be thorough by nature but may be over-reporting detail.
   • Recommendation: Consider summarizing analysis output rather than full verbose reporting.

Optimization Opportunities

1. Incremental scanning for quality check workflows (Syntax Error, Compiler Quality, Malicious Code Scan)

   • Affected Workflows: Daily Syntax Error Quality Check, Daily Compiler Quality Check, Daily Malicious Code Scan Agent
   • Potential Savings: ~30-50% token reduction by scoping to changed files only
   • Action: Add git diff --name-only HEAD~1 scoping to limit analysis to recently modified files

2. Output verbosity controls for research agents

   • Affected Workflows: Copilot CLI Deep Research Agent, Copilot PR Conversation NLP Analysis, Ubuntu Actions Image Analyzer
   • Potential Savings: ~20-40% by adding explicit output length constraints to prompts
   • Action: Add max_output_tokens guidance or structured output formats to workflow prompts

3. Issue Monster token efficiency — already excellent

   • 23 runs averaging only 92,476 tokens/run — a model of efficient high-volume workflow design
   • Recommendation: Use Issue Monster's architecture as a reference for other high-frequency workflows

4. Contribution Check batching

   • 6 runs averaging 684K tokens each (4.1M total)
   • Action: Investigate whether multiple runs process the same PRs or if there's overlap; batching could reduce total runs

Workflow Health Summary

Category	Count
Successful runs	75 (of 96 total)
Runs without token data	21 (22%)
Total errors across all runs	36
Missing data events	4
Safe output items produced	79

Runs missing token data (by workflow):

• Issue Monster: 16 runs (likely very fast/lightweight runs that completed before metrics were captured)
• PR Triage Agent: 2 runs
• Daily Copilot Token Consumption Report: 1 run (current run)
• Documentation Noob Tester: 1 run
• Super Linter Report: 1 run

Historical Tracking Note

This is the first run of the token consumption report with historical tracking enabled. Today's aggregated metrics have been stored in repo memory at memory/token-metrics/history.jsonl (branch: memory/token-metrics).

Future reports will include:

• Day-over-day comparisons
• 7-day moving averages
• Week-over-week trend analysis
• Long-term cost projections (once cost data is available in logs)

Cost Data: The current log format reports total_cost: 0 in the summary. Once cost data becomes available in the logs, it will be incorporated into future reports automatically.

---

References: §23340261754

AI generated by Daily Copilot Token Consumption Report · history

• expires on Mar 23, 2026, 11:30 AM UTC

[github-actions[bot]]

This discussion has been marked as outdated by Daily Copilot Token Consumption Report.

A newer discussion is available at Discussion #22415.