Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jonwiggins/optio/llms.txt

Use this file to discover all available pages before exploring further.

Optio records the USD cost of every agent run — both coding and review tasks — extracted directly from the agent’s output. The Costs dashboard (/costs) aggregates this data so you can understand where spend is going and catch runaway tasks early.

​
The Costs dashboard

Open the dashboard from the sidebar (dollar sign icon) or navigate to /costs.

​
Summary cards

The top row shows four summary metrics for the selected period:
MetricDescription
Total costSum of all task costs in USD
Task countNumber of tasks with recorded cost
Average costMean cost per task
Cost trendPercentage change vs. the previous equivalent period

​
Charts

Daily costs (area chart) — per-day cost and task count for the selected period. Use this to spot unusual spikes. Cost by repository (bar chart) — total spend aggregated per repository, sorted by highest cost. Useful for identifying which repos are driving the most spend. Cost by task type (pie chart) — breakdown between coding tasks and review tasks.

​
Top 10 most expensive tasks

A table listing the ten costliest tasks in the period, with columns for title, repository, task type, state, model used, input/output tokens, and cost in USD. Click any task to open its detail view.

​
Filtering

Two filters are available at the top of the dashboard:
  • Period — select 7d, 14d, 30d, or 90d. Defaults to 30 days.
  • Repository — filter all charts and the top-tasks table to a single repository.
Changing either filter immediately refreshes all charts and summary cards.

​
Programmatic access

The same data is available via the API: 
GET /api/analytics/costs
Query parameters:
ParameterTypeDefaultDescription
daysinteger30Number of days to include
repoUrlstring—Filter to a specific repository URL
Response shape: 
{
  "summary": {
    "totalCost": "4.2310",
    "taskCount": 87,
    "tasksWithCost": 83,
    "avgCost": "0.0509",
    "costTrend": "-12.4",
    "prevPeriodCost": "4.8200",
    "days": 30
  },
  "forecast": {
    "dailyAvgCost": "0.1410",
    "monthCostSoFar": "2.1800",
    "forecastedMonthTotal": "5.0900",
    "daysRemaining": 14
  },
  "dailyCosts": [
    { "date": "2025-01-01", "cost": 0.32, "taskCount": 6 }
  ],
  "costByRepo": [
    { "repoUrl": "https://github.com/org/repo", "totalCost": 1.84, "taskCount": 32 }
  ],
  "costByType": [
    { "taskType": "coding", "totalCost": 3.91, "taskCount": 74 },
    { "taskType": "review", "totalCost": 0.32, "taskCount": 9 }
  ],
  "costByModel": [
    {
      "model": "claude-opus-4-20250514",
      "totalCost": 3.12,
      "taskCount": 42,
      "successRate": 95,
      "avgCost": 0.074,
      "totalInputTokens": 5880000,
      "totalOutputTokens": 756000
    }
  ],
  "anomalies": [
    {
      "id": "uuid",
      "title": "Migrate database schema",
      "repoUrl": "https://github.com/org/repo",
      "taskType": "coding",
      "state": "completed",
      "costUsd": "4.2100",
      "modelUsed": "claude-opus-4-20250514",
      "repoAvgCost": 0.84,
      "costRatio": 5.01,
      "createdAt": "2025-01-12T09:00:00Z"
    }
  ],
  "modelSuggestions": [
    {
      "repoUrl": "https://github.com/org/repo",
      "currentModel": "claude-opus-4-20250514",
      "taskCount": 15,
      "avgCost": 0.92,
      "cheaperModelAvgCost": 0.18
    }
  ],
  "topTasks": [
    {
      "id": "uuid",
      "title": "Refactor auth module",
      "repoUrl": "https://github.com/org/repo",
      "taskType": "coding",
      "state": "completed",
      "costUsd": "0.8140",
      "inputTokens": 142000,
      "outputTokens": 18000,
      "modelUsed": "claude-opus-4-20250514",
      "createdAt": "2025-01-15T10:32:00Z"
    }
  ]
}
All cost values are returned as strings or numbers depending on the field. The summary cost fields are strings to preserve decimal precision.

​
Additional response fields

FieldDescription
costByModelCost aggregated by model, including success rate and token totals
anomaliesTasks costing 3× or more the repository average (up to 20 results)
modelSuggestionsRepos using expensive models where a cheaper model has proven results
Review tasks typically require far fewer turns than coding tasks. Set reviewModel to sonnet (the default) in your repository settings to use a cheaper model for review subtasks without sacrificing quality.
Set maxTurnsCoding on a repository to cap the maximum number of agent turns per coding task. Fewer turns means lower cost per task and a predictable spend ceiling. You can configure this in the repository settings under Agent settings.

Build docs developers (and LLMs) love

Get started for freeTalk to us