leo-claude-mktplace/docs/TEST_01_PROJMAN.md

Projman Plugin Testing Plan

Status: Phase 2 & 3 Complete - Ready for Testing Date: 2025-11-18 Plugin Version: 0.1.0

Overview

This document outlines the testing strategy for the Projman plugin, which has completed Phase 2 (Commands) and Phase 3 (Agents).

What Was Built

Phase 2: Commands (5 total)

• ✅ /sprint-plan - AI-guided planning with planner agent
• ✅ /sprint-start - Begin execution with orchestrator agent
• ✅ /sprint-status - Quick progress check
• ✅ /sprint-close - Capture lessons learned (critical!)
• ✅ /labels-sync - Sync label taxonomy from Gitea

Phase 3: Agents (3 total)

• ✅ Planner Agent - Thoughtful, asks clarifying questions, searches lessons learned
• ✅ Orchestrator Agent - Concise, action-oriented, tracks progress meticulously
• ✅ Executor Agent - Implementation-focused, follows specs precisely

Supporting Components

• ✅ Plugin manifest (plugin.json) with valid schema
• ✅ MCP configuration (.mcp.json) for Gitea + Wiki.js
• ✅ Label taxonomy skill with suggestion logic
• ✅ README.md with complete usage guide
• ✅ CONFIGURATION.md with step-by-step setup

Total: 13 files, ~3,719 lines of documentation

Testing Setup

Prerequisites Completed

✅ MCP Servers Installed:

• mcp-servers/gitea/.venv/ - Gitea MCP Server
• mcp-servers/wikijs/.venv/ - Wiki.js MCP Server

✅ System Configuration:

• ~/.config/claude/gitea.env - Gitea credentials
• ~/.config/claude/wikijs.env - Wiki.js credentials

✅ Project Configuration:

• .env - Project-specific settings (NOT committed)

  GITEA_REPO=claude-code-hhl-toolkit
  WIKIJS_PROJECT=projects/claude-code-hhl-toolkit
  

✅ Local Test Marketplace:

• .claude-plugins/projman-test-marketplace/marketplace.json
• Points to ../../projman for local testing

Repository Structure

hhl-claude-agents/
├── .env                              ✅ Project config (in .gitignore)
├── .claude-plugins/
│   └── projman-test-marketplace/
│       └── marketplace.json          ✅ Local marketplace
├── projman/                          ✅ Complete plugin
│   ├── .claude-plugin/
│   │   └── plugin.json
│   ├── .mcp.json
│   ├── commands/
│   │   ├── sprint-plan.md
│   │   ├── sprint-start.md
│   │   ├── sprint-status.md
│   │   ├── sprint-close.md
│   │   └── labels-sync.md
│   ├── agents/
│   │   ├── planner.md
│   │   ├── orchestrator.md
│   │   └── executor.md
│   ├── skills/
│   │   └── label-taxonomy/
│   │       └── labels-reference.md
│   ├── README.md
│   └── CONFIGURATION.md
└── mcp-servers/
    ├── gitea/
    │   └── .venv/
    └── wikijs/
        └── .venv/

Pre-Flight Checks

1. Verify MCP Server Connectivity

Test Gitea Connection:

cd mcp-servers/gitea
source .venv/bin/activate
python -c "from mcp_server.config import load_config; config = load_config(); print(f'✅ Gitea: {config.api_url}')"

Expected output:

✅ Gitea: http://gitea.hotport/api/v1

Test Wiki.js Connection:

cd mcp-servers/wikijs
source .venv/bin/activate
python -c "from mcp_server.config import load_config; config = load_config(); print(f'✅ Wiki.js: {config.api_url}')"

Expected output:

✅ Wiki.js: http://wikijs.hotport/graphql

2. Verify Configuration Files

Check System Config:

ls -la ~/.config/claude/*.env
# Should show:
# -rw------- gitea.env
# -rw------- wikijs.env

Check Project Config:

cat .env
# Should show:
# GITEA_REPO=claude-code-hhl-toolkit
# WIKIJS_PROJECT=projects/claude-code-hhl-toolkit

Verify .env is ignored:

git check-ignore .env
# Should output: .env

3. Verify Plugin Structure

Check plugin manifest:

cat projman/.claude-plugin/plugin.json | python3 -m json.tool > /dev/null && echo "✅ Valid JSON"

Check MCP config:

cat projman/.mcp.json | python3 -m json.tool > /dev/null && echo "✅ Valid JSON"

List all components:

tree projman/ -L 2

Testing Phases

Phase 1: Quick Validation (5-10 minutes)

Goal: Verify basic connectivity and command loading

Test 1.1: Label Sync (No agent, pure MCP test)

/labels-sync

Expected Behavior:

• ✅ Checks git branch first
• ✅ Connects to Gitea MCP server
• ✅ Fetches organization labels (28)
• ✅ Fetches repository labels (16)
• ✅ Shows total count (44 labels)
• ✅ Updates projman/skills/label-taxonomy/labels-reference.md
• ✅ Confirms successful sync

Success Criteria:

• No connection errors
• Label counts match Gitea
• File updated with current timestamp
• All label categories present (Agent, Complexity, Efforts, Priority, Risk, Source, Type, Component, Tech)

Test 1.2: Sprint Status (Read-only test)

/sprint-status

Expected Behavior:

• ✅ Checks git branch
• ✅ Fetches open issues from Gitea
• ✅ Fetches closed issues from Gitea
• ✅ Categorizes by status (Open, In Progress, Blocked, Completed)
• ✅ Shows completion percentage
• ✅ Identifies priority alerts

Success Criteria:

• Issues fetch successfully
• Categorization works
• No write operations attempted
• Progress summary accurate

Phase 2: Agent Validation (15-20 minutes)

Goal: Test agent personalities and MCP tool integration

Test 2.1: Planner Agent (via /sprint-plan)

/sprint-plan

Test Input:

"Plan a small sprint to add usage examples to the projman README"

Expected Planner Behavior:

1. ✅ Checks git branch (development)
2. ✅ Asks clarifying questions:
   • What kind of examples?
   • How detailed should they be?
   • Any specific use cases?
3. ✅ Searches lessons learned:
   • Uses search_lessons MCP tool
   • Searches by tags: "documentation", "readme"
4. ✅ Performs architecture analysis:
   • Thinks through structure
   • Considers edge cases
   • References past lessons
5. ✅ Creates Gitea issues:
   • Uses suggest_labels for each issue
   • Creates 2-3 well-structured issues
   • Includes acceptance criteria
   • References architectural decisions
6. ✅ Generates planning document:
   • Summarizes sprint goals
   • Lists created issues
   • Documents assumptions

Success Criteria:

• Planner personality evident (thoughtful, asks questions)
• Lessons learned searched proactively
• Labels suggested intelligently
• Issues created in Gitea with proper structure
• Architecture analysis thorough

Test 2.2: Orchestrator Agent (via /sprint-start)

/sprint-start

Expected Orchestrator Behavior:

1. ✅ Checks git branch
2. ✅ Fetches sprint issues from Gitea
3. ✅ Searches relevant lessons:
   • Uses search_lessons with tags
   • Presents relevant past experiences
4. ✅ Identifies next task:
   • Highest priority
   • Unblocked by dependencies
5. ✅ Generates lean execution prompt:
   • Concise (not verbose)
   • Actionable steps
   • References lessons
   • Clear acceptance criteria

Success Criteria:

• Orchestrator personality evident (concise, action-oriented)
• Lessons searched by relevant tags
• Next task identified correctly
• Execution prompt is lean (not planning document)
• Dependencies checked

Test 2.3: Executor Agent (Manual invocation if needed)

Note: Executor typically invoked by orchestrator, but can be tested independently.

Expected Executor Behavior:

1. ✅ Checks git branch
2. ✅ Follows specifications precisely
3. ✅ Writes clean, tested code
4. ✅ Handles edge cases
5. ✅ References lessons learned in code comments
6. ✅ Generates completion report

Success Criteria:

• Executor personality evident (implementation-focused)
• Code follows specs exactly
• Tests included
• Edge cases covered
• Lessons applied in implementation

Phase 3: Full Workflow Test (30-45 minutes)

Goal: Complete sprint lifecycle end-to-end

Scenario: "Add comprehensive testing examples to projman documentation"

Step 3.1: Planning (/sprint-plan)

/sprint-plan

Input: "Add comprehensive testing examples to projman documentation,
including command usage, agent behavior, and troubleshooting scenarios"

Expected Flow:

1. Planner asks clarifying questions
2. Searches lessons about documentation
3. Creates 3-4 issues in Gitea:
   • Add command usage examples
   • Add agent behavior examples
   • Add troubleshooting guide
   • Add quick start tutorial
4. Suggests appropriate labels for each

Validation:

• Check Gitea - issues created?
• Check labels - appropriate categories?
• Check issue bodies - acceptance criteria clear?

Step 3.2: Execution (/sprint-start)

/sprint-start

Expected Flow:

1. Orchestrator reviews issues
2. Searches lessons about documentation
3. Identifies first task
4. Generates lean execution prompt

Validation:

• Next task correctly identified?
• Execution prompt concise?
• Lessons referenced?

Step 3.3: Work on Task

Implement the first task (e.g., add command examples to README).

Step 3.4: Close Sprint (/sprint-close)

/sprint-close

Expected Flow:

1. Orchestrator reviews completion
2. Asks questions about sprint:
   • What challenges faced?
   • What went well?
   • Preventable mistakes?
3. Captures lessons learned:
   • Structures in proper format
   • Suggests appropriate tags
4. Saves to Wiki.js:
   • Uses create_lesson MCP tool
   • Creates in /projects/claude-code-hhl-toolkit/lessons-learned/sprints/
5. Offers git operations:
   • Commit changes
   • Merge branches
   • Tag sprint

Validation:

• Lessons captured in proper format?
• Saved to Wiki.js successfully?
• Tags appropriate for discovery?
• Check Wiki.js - lesson visible?

Phase 4: Edge Case Testing (15-20 minutes)

Goal: Test branch detection and security

Test 4.1: Production Branch Detection

git checkout main  # Switch to production
/sprint-plan

Expected Behavior:

• ❌ Command blocks immediately
• ❌ Shows production branch warning
• ❌ Instructs user to switch to development
• ❌ Does NOT proceed with planning

Test 4.2: Staging Branch Detection

git checkout -b staging  # Create staging branch
/sprint-start

Expected Behavior:

• ⚠️ Command warns about staging
• ⚠️ Limited capabilities (can create issues, cannot modify code)
• ⚠️ Instructs to switch to development for execution

Test 4.3: Development Branch (Normal)

git checkout development  # Back to development
/sprint-plan

Expected Behavior:

• ✅ Full capabilities enabled
• ✅ No warnings
• ✅ Normal operation

Validation:

• Production branch blocked?
• Staging branch limited?
• Development branch full access?

Phase 5: Error Handling (10-15 minutes)

Goal: Test graceful error handling

Test 5.1: Invalid Configuration

Temporarily rename .env:

mv .env .env.bak
/sprint-status

Expected Behavior:

• ❌ Clear error message about missing configuration
• ❌ Instructions to create .env
• ❌ No cryptic errors

Test 5.2: Network Issues (Simulate)

Stop Gitea or Wiki.js service temporarily:

/labels-sync

Expected Behavior:

• ❌ Connection error clearly stated
• ❌ Helpful troubleshooting suggestions
• ❌ No crashes or stack traces

Test 5.3: Invalid Repository

Edit .env with wrong repo name:

echo "GITEA_REPO=nonexistent-repo" > .env
/sprint-status

Expected Behavior:

• ❌ Repository not found error
• ❌ Suggestions to check .env configuration
• ❌ No silent failures

Cleanup:

mv .env.bak .env  # Restore configuration

Success Metrics

Technical Metrics

• All MCP servers connect successfully
• All 5 commands execute without errors
• All 3 agents exhibit correct personalities
• Branch detection works 100% accurately
• Labels sync correctly from Gitea
• Issues created with proper structure and labels
• Lessons learned saved to Wiki.js successfully
• No hardcoded secrets or absolute paths
• Error messages clear and actionable

User Experience Metrics

• Commands intuitive to use
• Agent personalities distinct and helpful
• Planner asks relevant questions
• Orchestrator provides concise guidance
• Executor follows specs precisely
• Error messages helpful (not cryptic)
• Documentation clear and accurate

Quality Metrics

• No crashes or unhandled exceptions
• Branch security enforced correctly
• Configuration validation works
• MCP tool integration seamless
• Label suggestions intelligent
• Lessons learned captured systematically

Known Limitations (Phase 0.1.0)

1. No Executor Integration - Executor agent not yet invoked automatically by orchestrator (Phase 4)
2. No Milestone Support - Sprint milestones not implemented (Phase 4)
3. No Dependencies Tracking - Issue dependencies not automatically tracked (Phase 4)
4. No Progress Updates - Orchestrator doesn't automatically update issue comments (Phase 4)
5. Manual Git Operations - Git operations not automated yet (Phase 4)

These are expected at this stage and will be addressed in Phase 4 (Lessons Learned Integration).

Troubleshooting Guide

Issue: Commands not found

Symptoms: /sprint-plan returns "command not found"

Solutions:

1. Check marketplace loaded: ls .claude-plugins/projman-test-marketplace/
2. Verify plugin path in marketplace.json
3. Restart Claude Code

Issue: MCP connection errors

Symptoms: "Failed to connect to Gitea" or "Failed to connect to Wiki.js"

Solutions:

1. Check system config exists: ls ~/.config/claude/*.env
2. Verify API URLs correct in config files
3. Test MCP servers manually (see Pre-Flight Checks)
4. Check network connectivity to services

Issue: Repository not found

Symptoms: "Repository 'X' not found in organization"

Solutions:

1. Check .env file: cat .env
2. Verify GITEA_REPO matches actual repository name
3. Check Gitea organization matches GITEA_OWNER in system config
4. Verify API token has access to repository

Issue: Lessons not saving to Wiki.js

Symptoms: "Failed to create lesson" or permission errors

Solutions:

1. Check Wiki.js API token has Pages (create) permission
2. Verify WIKIJS_BASE_PATH exists in Wiki.js
3. Check WIKIJS_PROJECT path is correct
4. Test Wiki.js connection (see Pre-Flight Checks)

Issue: Branch detection not working

Symptoms: Can create issues on production branch

Solutions:

1. Verify git repository initialized: git status
2. Check branch name: git branch --show-current
3. Review agent prompts - branch check should be first operation
4. This is a critical bug - report immediately

Next Steps After Testing

If All Tests Pass ✅

1. Document Findings

   • Create test report with results
   • Note any minor issues encountered
   • Capture user experience feedback

2. Move to Phase 4: Lessons Learned Integration

   • Implement automatic issue updates
   • Add milestone support
   • Implement dependency tracking
   • Automate git operations

3. Prepare for Phase 5: Testing & Validation

   • Write integration tests
   • Test with real sprint on CuisineFlow
   • Collect user feedback from team

If Tests Fail ❌

1. Document Failures

   • Exact error messages
   • Steps to reproduce
   • Expected vs actual behavior

2. Categorize Issues

   • Critical: Blocks basic functionality
   • High: Major feature doesn't work
   • Medium: Feature works but has issues
   • Low: Minor UX improvements

3. Fix and Retest

   • Fix critical issues first
   • Retest after each fix
   • Update documentation if needed

Test Execution Log

Test Session 1: [Date]

Tester: [Name] Duration: [Time] Environment:

• Branch: [branch name]
• Claude Code Version: [version]
• Plugin Version: 0.1.0

Results:

Test	Status	Notes
Pre-Flight: MCP Connectivity	[ ] Pass / [ ] Fail
Pre-Flight: Configuration	[ ] Pass / [ ] Fail
1.1: Label Sync	[ ] Pass / [ ] Fail
1.2: Sprint Status	[ ] Pass / [ ] Fail
2.1: Planner Agent	[ ] Pass / [ ] Fail
2.2: Orchestrator Agent	[ ] Pass / [ ] Fail
2.3: Executor Agent	[ ] Pass / [ ] Fail
3: Full Workflow	[ ] Pass / [ ] Fail
4: Branch Detection	[ ] Pass / [ ] Fail
5: Error Handling	[ ] Pass / [ ] Fail

Overall Assessment: [ ] Pass / [ ] Fail

Critical Issues Found: [Number]

Recommendations: [Next steps]

---

Testing Status: Ready to Begin Next Step: Execute Pre-Flight Checks and Phase 1 Quick Validation