Skip to content

An AI-powered task-management system you can drop into Cursor, Lovable, Windsurf, Roo, and others.

2.6 KB
Raw Blame

Task Master Progress Testing Guide

Quick reference for testing streaming/non-streaming functionality with token tracking.

🎯 Test Modes

  1. MCP Streaming - Has reportProgress + mcpLog, shows emoji indicators (πŸ”΄πŸŸ πŸŸ’)
  2. CLI Streaming - No reportProgress, shows terminal progress bars
  3. Non-Streaming - No progress reporting, single response

πŸš€ Quick Commands

# Test Scripts (accept: mcp-streaming, cli-streaming, non-streaming, both, all)
node test-parse-prd.js [mode] 
node test-analyze-complexity.js [mode]
node test-expand.js [mode] [num_subtasks]
node test-expand-all.js [mode] [num_subtasks]
node parse-prd-analysis.js [accuracy|complexity|all]

# CLI Commands
node scripts/dev.js parse-prd test.txt         # Local dev (streaming)
node scripts/dev.js analyze-complexity --research
node scripts/dev.js expand --id=1 --force
node scripts/dev.js expand --all --force

task-master [command]                          # Global CLI (non-streaming)

βœ… Success Indicators

Indicators

  • Priority: πŸ”΄πŸ”΄πŸ”΄ (high), 🟠🟠βšͺ (medium), 🟒βšͺβšͺ (low)
  • Complexity: ●●● (7-10), ●●○ (4-6), ●○○ (1-3)

Token Format

Tokens (I/O): 2,150/1,847 ($0.0423) (~4 chars per token)

Progress Bars

Single:  Generating subtasks... |β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‘β–‘| 80% (4/5)
Dual:    Expanding 3 tasks | Task 2/3 |β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‘β–‘| 66%
         Generating 5 subtasks... |β–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–ˆβ–‘β–‘β–‘β–‘| 60%

Fractional Progress

(completedTasks + currentSubtask/totalSubtasks) / totalTasks Example: 33% β†’ 46% β†’ 60% β†’ 66% β†’ 80% β†’ 93% β†’ 100%

πŸ› Quick Fixes

Issue Fix
No streaming Check reportProgress is passed
NaN% progress Filter duplicate subtask_progress events
Missing tokens Check .env has API keys
Broken bars Terminal width > 80
projectRoot.split Use projectRoot not session
# Debug
TASKMASTER_DEBUG=true node test-expand.js
npm run lint

πŸ“Š Benchmarks

  • Single task: 10-20s (5 subtasks)
  • Expand all: 30-45s (3 tasks)
  • Streaming: ~10-20% faster
  • Updates: Every 2-5s

πŸ”„ Test Workflow

# Quick check
node test-parse-prd.js both && npm test

# Full suite (before release)
for test in parse-prd analyze-complexity expand expand-all; do
  node test-$test.js all
done
node parse-prd-analysis.js all
npm test

🎯 MCP Tool Example

{
  "tool": "parse_prd",
  "args": {
    "input": "prd.txt",
    "numTasks": "8", 
    "force": true,
    "projectRoot": "/path/to/project"
  }
}