Skip to content

📊 Copilot Token Usage Report2026-04-29 #2289

@github-actions

Description

@github-actions

Overview

Period: 2026-04-28 to 2026-04-29
Runs analyzed: 50 (49 had token data)
Total tokens: 11,139,893 across all workflows
Estimated total cost: $0.00 (cost tracking not yet enabled — tokens tracked via AWF api-proxy)
Total Actions minutes: 332 min

All runs used model: claude-sonnet-4.6 (via Copilot provider). Average cache efficiency: ~44%.

Workflow Summary

Workflow Runs Total Tokens Avg Tokens/Run Avg Cache Eff Action Min Allocation
Build Test Suite 9 3,535K 393K 46% high
Firewall Issue Dispatcher 3 1,743K 581K 48% high
Smoke Copilot BYOK 11 1,300K 118K 43% high
Smoke Copilot 9 1,134K 126K 43% medium
Smoke Services 3 821K 274K 47% medium
Daily Claude Token Usage Analyzer 1 529K 529K 48% medium
Smoke Chroot 6 408K 68K 40% low
CI/CD Pipelines Gap Assessment 1 407K 407K 44% one-off
Issue Duplication Detector 3 362K 121K 41% low
Daily Security Review 1 344K 344K 45% daily
Dependency Security Monitor 1 342K 342K 47% daily
Pelis Agent Factory Advisor 1 215K 215K 44% one-off
Daily Copilot Token Usage Analyzer 1 0 0 (this run)

🔍 Optimization Opportunities

  1. Firewall Issue Dispatcher — 581K tokens/run (⚠️ >100K threshold)

    • Highest per-run token consumption of any recurring workflow
    • One run hit 886K tokens (2026-04-29) — 2.2× its usual average
    • Cache efficiency is healthy at 48%, but absolute input volume is high
    • Consider reviewing system prompt length and number of MCP tool calls per dispatch
  2. Build Test Suite — 393K tokens/run (⚠️ >100K threshold)

    • One run spiked to 829K tokens (run §25072868929) — 2.1× the average
    • Highly variable: range 124K–829K suggests non-deterministic work or retry loops
    • Review whether test failures cause expensive retry or error-recovery turns
  3. Daily Claude Token Usage Analyzer — 529K tokens/run (⚠️ >100K threshold)

    • Single daily run but high consumption; consider reducing report verbosity or data scope
  4. CI/CD Pipelines Gap Assessment — 407K tokens/run (⚠️ >100K threshold)

    • One-off assessment; acceptable as a one-time cost but flag if recurring
  5. ⚠️ Total tokens up ~20% vs previous report (11.1M vs 9.3M yesterday)

    • Primarily driven by Build Test Suite (3.5M, up from ~2.7M) and Firewall Issue Dispatcher (1.7M, new high)
    • Warrants monitoring over next 2–3 days to determine if trend continues
Per-Workflow Run Details

Build Test Suite (9 runs)

Firewall Issue Dispatcher (3 runs)

Smoke Copilot BYOK (11 runs)

Smoke Copilot (9 runs)

Smoke Services (3 runs)

Smoke Chroot (6 runs)

One-off / Daily Workflows

  • Daily Claude Token Usage Analyzer: 529K — §25075334909
  • CI/CD Pipelines Gap Assessment: 407K — §25054360522
  • Daily Security Review: 344K — §25055087899
  • Dependency Security Monitor: 342K — §25083205929
  • Issue Duplication Detector: 362K total (3 runs) — avg 121K
  • Pelis Agent Factory Advisor: 215K — §25079367915
Workflows Without Token Data

Runs where token_usage was null or 0 — these may not have fully initialized before logging:

  • Daily Copilot Token Usage Analyzer (this run) — §25104566063 — token data not yet available at report generation time

Historical Trend

Date Total Tokens Runs Change
2026-04-29 11,139,893 50 ⬆️ +19.9%
2026-04-28 9,287,181 50

The ~20% increase is driven primarily by more Build Test Suite runs with higher individual counts and an elevated Firewall Issue Dispatcher run (886K). Smoke tests remained stable. Monitor for 2–3 more days before acting.

Previous Report

📊 Copilot Token Usage Report 2026-04-28

References:

Generated by Daily Copilot Token Usage Analyzer · ● 431.1K ·

Metadata

Metadata

Assignees

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions