📊 Copilot Token Usage Report2026-04-15

[github-actions]

Overview

Period: 2026-04-15 (UTC)
Runs analyzed: 50 total (44 with token data; 5 cancelled without data; 1 in-progress)
Total tokens: 14.5M (14,498,837) | Total effective tokens: 17.5M (17,482,782)
Estimated total cost: N/A (estimated_cost not populated by api-proxy)
Total Actions minutes: 328 min
Total turns: 44 across all runs
Total errors: 16 (across 13 workflows)

i️ All inference is via Copilot (model: claude-sonnet-4.6). Cost is not available — token counts are from AWF api-proxy firewall logs.

---

Workflow Summary

Workflow	Runs	Total Tokens	Avg Tokens/Run	Eff Tokens	Cache Eff	Errors
Build Test Suite	8	4,344K	543K	4,854K	88%	2/8 (25%)
Smoke Services	9	2,096K	233K	2,311K	87%	2/9 (22%)
Firewall Issue Dispatcher	3	1,786K	595K	1,996K	93%	1/3 (33%) ⚠️
Smoke Copilot	11	1,386K	126K	1,534K	78%	4/11 (36%) ⚠️
Daily Security Review and Threat Modeling	1	923K	923K	1,046K	89%	0/1
Dependency Security Monitor	1	880K	880K	983K	93%	1/1 (100%) ❌
Smoke OpenCode	5	869K	174K	954K	83%	2/5 (40%) ⚠️
Issue Duplication Detector	3	760K	253K	844K	83%	0/3
Daily Claude Token Optimization Advisor	1	0*	0*	1,330K	89%	1/1 (100%) ❌
Pelis Agent Factory Advisor	1	598K	598K	675K	86%	0/1
Daily Claude Token Usage Analyzer	1	473K	473K	540K	89%	0/1
Smoke Chroot	5	383K	77K	416K	66%	3/5 (60%) ⚠️
Daily Copilot Token Usage Analyzer	1	0*	—	—	—	0/1
Totals	50	14,498K	290K	17,483K	87%	16 errors

* token_usage is null/0 for Daily Claude Token Optimization Advisor (failed) and Daily Copilot Token Usage Analyzer (this run, in-progress). The Optimizer has firewall-level effective tokens indicating it consumed ~1.3M tokens before failing.

---

🔍 Optimization Opportunities

1. Build Test Suite — 543K avg tokens/run (+22.6% vs yesterday ⚠️ exceeds 20% trend threshold)

   • 7 successful runs consumed 4.3M total tokens in a single day
   • Average up from 443K/run yesterday — investigate what changed (new tests, larger context, more MCP tool calls)
   • 88% cache efficiency is healthy; focus on why the base prompt/context is growing

2. Dependency Security Monitor — 880K tokens/run, failed run (+303% vs yesterday's 218K avg ⚠️)

   • Single run failed (conclusion: failure) after consuming 880K tokens — wasted inference cost
   • Investigate failure root cause before next run to prevent recurrence
   • Previous average was 218K; this run was 4× higher, suggesting the failure path explored more context

3. Firewall Issue Dispatcher — 595K avg/run, one anomalous 1.3M-token failure

   • Run §24455120367 consumed 1.3M tokens before failing — ~2× the normal 245K avg for successful runs
   • Add early-exit guardrails or token budget limits to prevent runaway failure runs
   • 33% error rate (1/3 runs) warrants investigation

4. Daily Security Review and Threat Modeling — 923K tokens/run (9× the 100K threshold)

   • Highest single-run consumption today (down from 1.83M yesterday — improvement!)
   • Consider scoping reviews to diff-only or changed files when no major changes land

5. Smoke Chroot — 60% error rate (3/5 runs had errors) ⚠️

   • Errors do not translate to failures (all 5 runs completed successfully) but logged errors indicate instability
   • Lowest cache efficiency among smoke tests (66%) — review prompt stability

6. Daily Claude Token Optimization Advisor — 100% failure rate, 1.33M effective tokens consumed

   • Run failed but consumed significant tokens before aborting
   • Fix the failure root cause; this workflow should not be spending 1.3M tokens then failing

---

Per-Workflow Run Details

Build Test Suite (8 runs)

Run	Tokens	Conclusion
§24477096722	828K	✅ success
§24470583755	692K	✅ success
§24469880156	627K	✅ success
§24467006844	635K	✅ success
§24466516274	582K	✅ success
§24476706304	509K	✅ success
§24436400446	470K	✅ success
§24435896709	0	🚫 cancelled

Firewall Issue Dispatcher (3 runs)

Run	Tokens	Conclusion
§24455120367	1,301K ⚠️	❌ failure
§24471993687	274K	✅ success
§24440943135	212K	✅ success

Smoke Copilot (11 runs)

Run	Tokens	Conclusion
§24477096752	178K	✅ success
§24470576010	147K	✅ success
§24471612919	147K	✅ success
§24469880171	176K	✅ success
§24466516327	176K	✅ success
§24455351338	180K	❌ failure
§24476706320	89K	✅ success
§24467006852	88K	✅ success
§24436400429	89K	✅ success
§24435902129	116K	✅ success
§24472515865	0	🚫 cancelled

Runs Without Token Data

These runs had token_usage = null/0. They may have been cancelled, failed before inference, or the api-proxy was not enabled:

Workflow	Run	Conclusion	Notes
Daily Copilot Token Usage Analyzer	§24482607097	in_progress	This run
Daily Claude Token Optimization Advisor	§24455856010	❌ failure	Has firewall eff_tokens=1.33M
Smoke Copilot	§24472515865	🚫 cancelled	—
Smoke Services	§24472515891	🚫 cancelled	—
Smoke Chroot	§24472515925	🚫 cancelled	—
Smoke OpenCode	§24472515915	🚫 cancelled	—
Build Test Suite	§24435896709	🚫 cancelled	—

---

Historical Trend

Metric	2026-04-14	2026-04-15	Δ
Total runs	42	50	+19%
Total tokens	12.7M	14.5M	+14.2%
Effective tokens	15.0M	17.5M	+16.7%
Action minutes	220 min	328 min	+49% ⚠️
Total turns	39	44	+13%
Avg tokens/run	302K	290K	-4%
Build Test Suite avg	443K	543K	+22.6% ⚠️
Dependency Security Monitor avg	218K	880K	+303% ⚠️
Daily Security Review avg	1,830K	923K	-49.6% ✅

Notable changes:

• 📈 Action minutes jumped +49% (220 → 328) — highest seen; driven by Daily Claude Token Optimization Advisor (67 min, failed) and increased Build Test Suite volume
• 📈 Build Test Suite avg tokens up +22.6% (crosses 20% trend threshold) — warrants investigation
• 📉 Daily Security Review improved -49.6% vs yesterday (1.83M → 923K)
• 📈 Dependency Security Monitor spiked 4× its previous average, and the run failed

Previous Report

📊 Copilot Token Usage Report 2026-04-14 (#1981)

---

References:

• §24482607097 — this report run
• §24477096722 — highest Build Test Suite run (828K tokens)
• §24455120367 — anomalous Firewall Issue Dispatcher failure (1.3M tokens)

Generated by Daily Copilot Token Usage Analyzer · ● 692.6K · ◷