Skip to content

TESTING.md

Latest commit

 

History

History
135 lines (105 loc) · 2.78 KB

TESTING.md

File metadata and controls

135 lines (105 loc) · 2.78 KB

Testing with Claude Pro / Max

A guide for testing AI Dev Team with Claude Pro ($20/mo) or Max ($100/mo) subscriptions.

Prerequisites

  1. Claude Code CLI installed and logged in:
npm install -g @anthropic-ai/claude-code
claude login
  1. Project initialized:
cd /path/to/ai-team
python scripts/init_project.py "Test Project" "A test project to validate the concept"

Rate Limits

Metric Pro ($20) Max ($100)
Messages/5h ~45 ~225
Recommended cycles/day 3-4 15-20
Agents per cycle 3 (minimal) 6-9 (full team)

Test Scenarios

1. Single Agent (smallest test)

# Run just the Product Owner
./scripts/orchestrator_cli.sh po

# Check output
cat board/backlog.md
cat board/inbox/pm.md

2. Minimal Team (3 agents)

# PO → Dev1 → Tester
./scripts/orchestrator_cli.sh minimal

# Check results
cat board/sprint.md
ls workspace/src/

3. Full Dev Team (6 agents)

# All core agents (recommended for Max plan)
./scripts/orchestrator_cli.sh dev-team

4. Full Team with Security (9 agents)

# Everyone — only with Max plan
./scripts/orchestrator_cli.sh all

Recommended Test Plan

Day 1: Validate the concept

# Morning — initialize project
python scripts/init_project.py "MyProject" "Project description"

# Test 1: PO defines the backlog
./scripts/orchestrator_cli.sh po
cat board/backlog.md

# Test 2: Minimal cycle
./scripts/orchestrator_cli.sh minimal

Day 2: Multiple cycles

# 3-4 minimal cycles throughout the day
./scripts/orchestrator_cli.sh minimal  # morning
./scripts/orchestrator_cli.sh minimal  # afternoon
./scripts/orchestrator_cli.sh minimal  # evening

# Track progress
cat board/sprint.md

Day 3: Evaluate

# What got done?
cat board/sprint.md | grep "DONE"

# What code was generated?
find workspace/src -name "*.py" | head -20

# Is agent communication working?
cat board/inbox/*.md

Cron for Automated Cycles

For Pro plan — max 3-4 cycles/day:

# Every 6 hours, work hours only
0 8,14,20 * * * /path/to/ai-team/scripts/orchestrator_cli.sh minimal >> /path/to/ai-team/logs/cron.log 2>&1

For Max plan — you can run more:

# Every 2 hours
0 */2 8-22 * * * /path/to/ai-team/scripts/orchestrator_cli.sh dev-team >> /path/to/ai-team/logs/cron.log 2>&1

Troubleshooting

"Rate limit reached"

# Wait for reset (~5 hours) or:
# - Use minimal mode (3 agents instead of 6-9)
# - Reduce cycle frequency

Claude Code not logged in

claude login
# Follow browser instructions

Agents produce poor results

# Check their inbox — they might lack context
cat board/inbox/dev1.md

# Check project.md — is it descriptive enough?
cat board/project.md
⚡