Workforce/Cost Dashboard

Cost Attribution

Total Spend
$799.50
-12.4%vs last month
Budget Remaining
$401
$800 of $1200 used
Cost per Task
$5.40
-8%vs last month
Token Usage
24.7M
tokens processed

Spend by Project

Click a bar to drill into agent breakdown

Novact OS$624.30
CC
Ma
Br
78.1% of total
OneComm v2$128.40
CC
Ma
Br
No
16.1% of total
MCP Gateway$32.80
CC
No
Ma
4.1% of total
Ops Bridge$14.00
CC
No
Ma
1.8% of total

Daily Spend — Last 30 Days

Top Cost Drivers This Month

#ActionAgentCost% Total
1Code generation
CC
Claude Code
$148.2018.5%
2Spec authoring
Ma
Maren
$78.209.8%
3File operations
CC
Claude Code
$72.409.1%
4Test execution
CC
Claude Code
$58.307.3%
5Pattern scanning
Br
Brue
$52.406.6%
6Verification checks
Ma
Maren
$41.505.2%
7Code generation
CC
Claude Code
$38.404.8%
8PR review
Br
Brue
$38.904.9%
9Git operations
CC
Claude Code
$38.104.8%
10Decision capture
No
Nova
$32.804.1%
11KB search
Ma
Maren
$32.104%
12Task routing
No
Nova
$28.403.6%
13Deployment
CC
Claude Code
$25.603.2%
14Claim verification
Br
Brue
$22.102.8%
15KB consolidation
Tu
Tuck
$18.902.4%

Budget Alerts

warningMCP Gateway

MCP Gateway trending 23% over monthly budget

At current daily spend rate ($2.52/day), MCP Gateway will exceed its $60/month budget by Apr 28. Primary driver: increased Claude Code deployment cycles during gateway hardening.

infoNovact OS

Maren's spec-authoring costs up 8% month-over-month

Spec authoring cost increased from $172.40 (March) to projected $186.40 (April, annualized). Investigating whether increased spec complexity or model pricing changes are the driver.

infoAll

Utility model savings: $312 this month

Routing 1,240 background tasks to GPT-4o-mini instead of Opus saved an estimated $312 in compute costs. Tuck's context pack compilations account for 64% of these savings.

Model Routing — Cost Comparison

ModelTasksTotal Cost$/1K input
Claude 4.6 Opus
Anthropic
186$436.80$0.0150
Claude 4.6 Sonnet
Anthropic
420$148.20$0.0030
GPT-5
OpenAI
98$124.80$0.0120
GPT-4o-mini
OpenAI
1240$18.40$0.0001
Gemini 2.5 Pro
Google
45$42.30$0.0070

Model Routing Policy

Route reasoning-heavy tasks (spec authoring, decision capture, complex code generation) to Claude 4.6 Opus.

Route high-volume routine tasks (context compilation, freshness audits, simple file operations) to Claude 4.6 Sonnet.

Route quality review tasks to GPT-5 for adversarial model diversity.

Fall back to GPT-4o-mini for background utility tasks.

Cost cap: $500/month per projectAlert at: 80% of budget