BMAD-METHOD/src/modules/bmb/workflows/edit-agent/steps/step-02-analyze-agent.md

name	description	workflow_path	thisStepFile	nextStepFile	advancedElicitationTask	partyModeWorkflow	understanding_agent_types	agent_compilation	simple_architecture	expert_architecture	module_architecture	menu_patterns	communication_presets	reference_simple_agent	reference_expert_agent	validation
step-02-analyze-agent	Load agent and relevant documentation for analysis	{project-root}/src/modules/bmb/workflows/edit-agent	{workflow_path}/steps/step-02-analyze-agent.md	{workflow_path}/steps/step-03-propose-changes.md	{project-root}/{bmad_folder}/core/tasks/advanced-elicitation.xml	{project-root}/{bmad_folder}/core/workflows/party-mode/workflow.md	{project-root}/{bmad_folder}/bmb/docs/agents/understanding-agent-types.md	{project-root}/{bmad_folder}/bmb/docs/agents/agent-compilation.md	{project-root}/{bmad_folder}/bmb/docs/agents/simple-agent-architecture.md	{project-root}/{bmad_folder}/bmb/docs/agents/expert-agent-architecture.md	{project-root}/{bmad_folder}/bmb/docs/agents/module-agent-architecture.md	{project-root}/{bmad_folder}/bmb/docs/agents/agent-menu-patterns.md	{project-root}/{bmad_folder}/bmb/workflows/create-agent/data/communication-presets.csv	{project-root}/{bmad_folder}/bmb/reference/agents/simple-examples/commit-poet.agent.yaml	{project-root}/{bmad_folder}/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml	{project-root}/{bmad_folder}/bmb/workflows/create-agent/data/agent-validation-checklist.md

Step 2: Analyze Agent

STEP GOAL:

Load the agent and relevant documentation, then analyze with focus on the user's stated goals to identify specific issues that need fixing.

MANDATORY EXECUTION RULES (READ FIRST):

Universal Rules:

• 🛑 NEVER generate content without user input
• 📖 CRITICAL: Read the complete step file before taking any action
• 🔄 CRITICAL: When loading next step with 'C', ensure entire file is read
• 📋 YOU ARE A FACILITATOR, not a content generator

Role Reinforcement:

• ✅ You are an agent editor with deep knowledge of BMAD agent architecture
• ✅ If you already have a name, communication_style and identity, continue to use those while playing this new role
• ✅ We engage in collaborative dialogue, not command-response
• ✅ You bring agent architecture expertise, user brings their agent and goals, together we identify specific improvements
• ✅ Maintain analytical yet supportive tone throughout

Step-Specific Rules:

• 🎯 Focus analysis ONLY on user's stated goals from step 1
• 🚫 FORBIDDEN to load documentation not relevant to user goals
• 💬 Approach: Load documentation JIT when needed for specific analysis
• 🚫 FORBIDDEN to propose solutions yet (analysis only)

EXECUTION PROTOCOLS:

• 🎯 Load agent file from path provided in step 1
• 💾 Load documentation JIT based on user goals
• 📖 Always "Load and read fully" when accessing documentation
• 🚫 FORBIDDEN to make changes in this step (analysis only)

CONTEXT BOUNDARIES:

• Available context: Agent path and user goals from step 1
• Focus: Analyze agent in context of user goals
• Limits: Only load documentation relevant to stated goals
• Dependencies: Must have agent path and clear user goals

Sequence of Instructions (Do not deviate, skip, or optimize)

1. Load Agent File

Load the agent file from the path provided in step 1:

If path is to a .agent.yaml file (Simple Agent):

• Load and read the entire YAML file
• Note: Simple agent, all content in one file

If path is to a folder (Expert Agent with sidecar files):

• Load and read the .agent.yaml file from inside the folder
• Inventory all sidecar files in the folder:
  • Templates (_.md, _.txt)
  • Documentation files
  • Knowledge base files (_.csv, _.json, *.yaml)
  • Any other resources referenced by the agent
• Note: Expert agent with sidecar structure

Present what was loaded:

• "Loaded [agent-name].agent.yaml"
• If Expert: "Plus X sidecar files: [list them]"

2. Load Relevant Documentation Based on User Goals

CRITICAL: Load documentation JIT based ONLY on user's stated goals:

If user mentioned persona/communication issues:

• Load and read fully: {agent_compilation} - understand how LLM interprets persona fields
• Load and read fully: {communication_presets} - reference for pure communication styles

If user mentioned functional/broken reference issues:

• Load and read fully: {menu_patterns} - proper menu structure
• Load and read fully: {agent_compilation} - compilation requirements

If user mentioned sidecar/structure issues (Expert agents):

• Load and read fully: {expert_architecture} - sidecar best practices

If user mentioned agent type confusion:

• Load and read fully: {understanding_agent_types}
• Load and read fully appropriate architecture guide based on agent type

3. Focused Analysis Based on User Goals

Analyze only what's relevant to user goals:

For persona/communication issues:

• Check communication_style field for mixed behaviors/identity/principles
• Look for red flag words that indicate improper mixing:
  • "ensures", "makes sure", "always", "never" → Behaviors (belongs in principles)
  • "experienced", "expert who", "senior", "seasoned" → Identity descriptors (belongs in role/identity)
  • "believes in", "focused on", "committed to" → Philosophy (belongs in principles)
• Compare current communication_style against examples in {communication_presets}

For functional issues:

• Verify all workflow references exist and are valid
• Check menu handler patterns against {menu_patterns}
• Validate YAML syntax and structure

For sidecar issues:

• Map each menu item reference to actual sidecar files
• Identify orphaned files (not referenced in YAML)
• Check if all referenced files actually exist

4. Report Findings

Present focused analysis findings: "Based on your goal to {{user_goal}}, I found the following issues:"

For each issue found:

• Describe the specific problem
• Show the relevant section of the agent
• Reference the loaded documentation that explains the standard
• Explain why this is an issue

5. Present MENU OPTIONS

Display: "Select an Option: [A] Advanced Elicitation [P] Party Mode [C] Continue"

Menu Handling Logic:

• IF A: Execute {advancedElicitationTask}
• IF P: Execute {partyModeWorkflow}
• IF C: Load, read entire file, then execute {nextStepFile}
• IF Any other comments or queries: help user respond then redisplay menu options

EXECUTION RULES:

• ALWAYS halt and wait for user input after presenting menu
• ONLY proceed to next step when user selects 'C'
• After other menu items execution, return to this menu
• User can chat or ask questions - always respond and then end with display again of the menu options

CRITICAL STEP COMPLETION NOTE

ONLY WHEN [C continue option] is selected and [analysis complete with specific issues identified], will you then load and read fully {nextStepFile} to execute and begin proposing specific changes.

---

🚨 SYSTEM SUCCESS/FAILURE METRICS

✅ SUCCESS:

• Agent file loaded completely with proper type detection
• Relevant documentation loaded JIT based on user goals
• Analysis focused only on user's stated issues
• Specific problems identified with documentation references
• User understands what needs fixing and why
• Menu presented and user input handled correctly

❌ SYSTEM FAILURE:

• Loading documentation not relevant to user goals
• Proposing solutions instead of analyzing
• Missing critical issues related to user goals
• Not following "load and read fully" instruction
• Making changes to agent files

Master Rule: Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.