mirror of
https://github.com/bmadcode/BMAD-METHOD.git
synced 2025-12-29 16:14:59 +00:00
feat: implement granular step-file workflow architecture with multi-menu support
## Major Features Added - **Step-file workflow architecture**: Transform monolithic workflows into granular step files for improved LLM adherence and consistency - **Multi-menu handler system**: New `handler-multi.xml` enables grouped menu items with fuzzy matching - **Workflow compliance checker**: Added automated compliance validation for all workflows - **Create/edit agent workflows**: New structured workflows for agent creation and editing ## Workflow Enhancements - **Create-workflow**: Expanded from 6 to 14 detailed steps covering tools, design, compliance - **Granular step execution**: Each workflow step now has dedicated files for focused execution - **New documentation**: Added CSV data standards, intent vs prescriptive spectrum, and common tools reference ## Complete Migration Status - **4 workflows fully migrated**: `create-agent`, `edit-agent`, `create-workflow`, and `edit-workflow` now use the new granular step-file architecture - **Legacy transformation**: `edit-workflow` includes built-in capability to transform legacy single-file workflows into the new improved granular format - **Future cleanup**: Legacy versions will be removed in a future commit after validation ## Schema Updates - **Multi-menu support**: Updated agent schema to support `triggers` array for grouped menu items - **Legacy compatibility**: Maintains backward compatibility with single `trigger` field - **Discussion enhancements**: Added conversational_knowledge recommendation for discussion agents ## File Structure Changes - Added: `create-agent/`, `edit-agent/`, `edit-workflow/`, `workflow-compliance-check/` workflows - Added: Documentation standards and CSV reference files - Refactored: `create-workflow/steps/` with detailed granular step files ## Handler Improvements - Enhanced `handler-exec.xml` with clearer execution instructions - Improved data passing context for executed files - Better error handling and user guidance This architectural change significantly improves workflow execution consistency across all LLM models by breaking complex processes into manageable, focused steps. The edit-workflow transformation tool ensures smooth migration of existing workflows to the new format.
This commit is contained in:
Brian Madison
@@ -0,0 +1,134 @@
|
||||
---
|
||||
name: 'step-01-discover-intent'
|
||||
description: 'Get agent path and user editing goals'
|
||||
|
||||
# Path Definitions
|
||||
workflow_path: '{project-root}/src/modules/bmb/workflows/edit-agent'
|
||||
|
||||
# File References
|
||||
thisStepFile: '{workflow_path}/steps/step-01-discover-intent.md'
|
||||
nextStepFile: '{workflow_path}/steps/step-02-analyze-agent.md'
|
||||
|
||||
# Task References
|
||||
advancedElicitationTask: '{project-root}/.bmad/core/tasks/advanced-elicitation.xml'
|
||||
partyModeWorkflow: '{project-root}/.bmad/core/workflows/party-mode/workflow.md'
|
||||
---
|
||||
|
||||
# Step 1: Discover Edit Intent
|
||||
|
||||
## STEP GOAL:
|
||||
|
||||
Get the agent path to edit and understand what the user wants to accomplish before proceeding to targeted analysis.
|
||||
|
||||
## 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 who helps users improve their BMAD agents
|
||||
- ✅ 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 improve the agent
|
||||
- ✅ Maintain collaborative guiding tone throughout
|
||||
|
||||
### Step-Specific Rules:
|
||||
|
||||
- 🎯 Focus only on getting agent path and understanding user goals
|
||||
- 🚫 FORBIDDEN to load any documentation or analyze the agent yet
|
||||
- 💬 Approach: Direct questions to understand what needs fixing
|
||||
- 🚫 FORBIDDEN to make suggestions or propose solutions
|
||||
|
||||
## EXECUTION PROTOCOLS:
|
||||
|
||||
- 🎯 Ask clear questions to get agent path and user goals
|
||||
- 💾 Store path and goals for next step
|
||||
- 📖 Do NOT load any references in this step
|
||||
- 🚫 FORBIDDEN to analyze agent content yet
|
||||
|
||||
## CONTEXT BOUNDARIES:
|
||||
|
||||
- Available context: User wants to edit an existing agent
|
||||
- Focus: Get path and understand goals ONLY
|
||||
- Limits: No analysis, no documentation loading, no suggestions
|
||||
- Dependencies: User must provide agent path
|
||||
|
||||
## Sequence of Instructions (Do not deviate, skip, or optimize)
|
||||
|
||||
### 1. Get Agent Path
|
||||
|
||||
Ask the user:
|
||||
"What agent do you want to edit? Please provide the path to:
|
||||
|
||||
- A .agent.yaml file (Simple agent)
|
||||
- A folder containing .agent.yaml (Expert agent with sidecar files)"
|
||||
|
||||
Wait for user response with the path.
|
||||
|
||||
### 2. Understand Editing Goals
|
||||
|
||||
Ask clear questions to understand what they want to accomplish:
|
||||
"What do you want to change about this agent?"
|
||||
|
||||
Listen for specific goals such as:
|
||||
|
||||
- Fix broken functionality
|
||||
- Update personality/communication style
|
||||
- Add or remove commands
|
||||
- Fix references or paths
|
||||
- Reorganize sidecar files (Expert agents)
|
||||
- Update for new standards
|
||||
|
||||
Continue asking clarifying questions until goals are clear.
|
||||
|
||||
### 3. Confirm Understanding
|
||||
|
||||
Summarize back to user:
|
||||
"So you want to edit the agent at {{agent_path}} to {{user_goals}}. Is that correct?"
|
||||
|
||||
### 4. 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 [agent path and goals obtained], will you then load and read fully `{nextStepFile}` to execute and begin agent analysis.
|
||||
|
||||
---
|
||||
|
||||
## 🚨 SYSTEM SUCCESS/FAILURE METRICS
|
||||
|
||||
### ✅ SUCCESS:
|
||||
|
||||
- Agent path clearly obtained and validated
|
||||
- User editing goals understood completely
|
||||
- User confirms understanding is correct
|
||||
- Menu presented and user input handled correctly
|
||||
|
||||
### ❌ SYSTEM FAILURE:
|
||||
|
||||
- Proceeding without agent path
|
||||
- Making suggestions or analyzing agent
|
||||
- Loading documentation in this step
|
||||
- Not confirming user goals clearly
|
||||
|
||||
**Master Rule:** Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.
|
||||
@@ -0,0 +1,202 @@
|
||||
---
|
||||
name: 'step-02-analyze-agent'
|
||||
description: 'Load agent and relevant documentation for analysis'
|
||||
|
||||
# Path Definitions
|
||||
workflow_path: '{project-root}/src/modules/bmb/workflows/edit-agent'
|
||||
|
||||
# File References
|
||||
thisStepFile: '{workflow_path}/steps/step-02-analyze-agent.md'
|
||||
nextStepFile: '{workflow_path}/steps/step-03-propose-changes.md'
|
||||
|
||||
# Task References
|
||||
advancedElicitationTask: '{project-root}/.bmad/core/tasks/advanced-elicitation.xml'
|
||||
partyModeWorkflow: '{project-root}/.bmad/core/workflows/party-mode/workflow.md'
|
||||
|
||||
# Documentation References (load JIT based on user goals)
|
||||
understanding_agent_types: '{project-root}/.bmad/bmb/docs/agents/understanding-agent-types.md'
|
||||
agent_compilation: '{project-root}/.bmad/bmb/docs/agents/agent-compilation.md'
|
||||
simple_architecture: '{project-root}/.bmad/bmb/docs/agents/simple-agent-architecture.md'
|
||||
expert_architecture: '{project-root}/.bmad/bmb/docs/agents/expert-agent-architecture.md'
|
||||
module_architecture: '{project-root}/.bmad/bmb/docs/agents/module-agent-architecture.md'
|
||||
menu_patterns: '{project-root}/.bmad/bmb/docs/agents/agent-menu-patterns.md'
|
||||
communication_presets: '{project-root}/.bmad/bmb/workflows/create-agent/data/communication-presets.csv'
|
||||
reference_simple_agent: '{project-root}/.bmad/bmb/reference/agents/simple-examples/commit-poet.agent.yaml'
|
||||
reference_expert_agent: '{project-root}/.bmad/bmb/reference/agents/expert-examples/journal-keeper/journal-keeper.agent.yaml'
|
||||
validation: '{project-root}/.bmad/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.
|
||||
@@ -0,0 +1,157 @@
|
||||
---
|
||||
name: 'step-03-propose-changes'
|
||||
description: 'Propose specific changes and get approval'
|
||||
|
||||
# Path Definitions
|
||||
workflow_path: '{project-root}/src/modules/bmb/workflows/edit-agent'
|
||||
|
||||
# File References
|
||||
thisStepFile: '{workflow_path}/steps/step-03-propose-changes.md'
|
||||
nextStepFile: '{workflow_path}/steps/step-04-apply-changes.md'
|
||||
agentFile: '{{agent_path}}'
|
||||
|
||||
# Task References
|
||||
advancedElicitationTask: '{project-root}/.bmad/core/tasks/advanced-elicitation.xml'
|
||||
partyModeWorkflow: '{project-root}/.bmad/core/workflows/party-mode/workflow.md'
|
||||
|
||||
# Documentation References (load JIT if needed)
|
||||
communication_presets: '{project-root}/.bmad/bmb/workflows/create-agent/data/communication-presets.csv'
|
||||
agent_compilation: '{project-root}/.bmad/bmb/docs/agents/agent-compilation.md'
|
||||
---
|
||||
|
||||
# Step 3: Propose Changes
|
||||
|
||||
## STEP GOAL:
|
||||
|
||||
Propose specific, targeted changes based on analysis and get user approval before applying them to the agent.
|
||||
|
||||
## 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 who helps users improve their BMAD agents through targeted changes
|
||||
- ✅ 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 improve the agent
|
||||
- ✅ Maintain collaborative guiding tone throughout
|
||||
|
||||
### Step-Specific Rules:
|
||||
|
||||
- 🎯 Focus only on proposing changes based on analysis from step 2
|
||||
- 🚫 FORBIDDEN to apply changes without explicit user approval
|
||||
- 💬 Approach: Present one change at a time with clear before/after comparison
|
||||
- 📋 Load references JIT when explaining rationale or providing examples
|
||||
|
||||
## EXECUTION PROTOCOLS:
|
||||
|
||||
- 🎯 Propose one change at a time with clear before/after comparison
|
||||
- 💾 Track approved changes for application in next step
|
||||
- 📖 Load references JIT if needed for examples or best practices
|
||||
- 🚫 FORBIDDEN to apply changes without explicit user approval
|
||||
|
||||
## CONTEXT BOUNDARIES:
|
||||
|
||||
- Available context: Analysis results from step 2, agent path, and user goals from step 1
|
||||
- Focus: Propose specific changes based on analysis, not apply them
|
||||
- Limits: Only propose changes, do not modify any files yet
|
||||
- Dependencies: Must have completed step 2 analysis results
|
||||
|
||||
## Sequence of Instructions (Do not deviate, skip, or optimize)
|
||||
|
||||
### 1. Present First Change
|
||||
|
||||
Based on analysis from step 2, propose the most important change first:
|
||||
|
||||
"I recommend fixing {{issue}} because {{reason}}.
|
||||
|
||||
**Current:**
|
||||
|
||||
```yaml
|
||||
{ { current_code } }
|
||||
```
|
||||
|
||||
**Proposed:**
|
||||
|
||||
```yaml
|
||||
{ { proposed_code } }
|
||||
```
|
||||
|
||||
This will help with {{benefit}}."
|
||||
|
||||
### 2. Explain Rationale
|
||||
|
||||
- Why this change matters for the agent's functionality
|
||||
- How it aligns with BMAD agent best practices
|
||||
- Reference loaded documentation if helpful for explaining
|
||||
|
||||
### 3. Load References if Needed
|
||||
|
||||
**Load references JIT when explaining:**
|
||||
|
||||
- If proposing persona changes: Load and read `{communication_presets}` for examples
|
||||
- If proposing structural changes: Load and read `{agent_compilation}` for requirements
|
||||
|
||||
### 4. Get User Approval
|
||||
|
||||
"Does this change look good? Should I apply it?"
|
||||
Wait for explicit user approval before proceeding.
|
||||
|
||||
### 5. Repeat for Each Issue
|
||||
|
||||
Go through each identified issue from step 2 analysis one by one:
|
||||
|
||||
- Present change with before/after
|
||||
- Explain rationale with loaded references if needed
|
||||
- Get explicit user approval for each change
|
||||
- Track which changes are approved vs rejected
|
||||
|
||||
### 6. 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: Save approved changes list to context, then only then load, read entire file, then execute {nextStepFile}
|
||||
- IF Any other comments or queries: help user respond then [Redisplay Menu Options](#6-present-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 [all proposed changes reviewed and user approvals obtained], will you then load and read fully `{nextStepFile}` to execute and begin applying approved changes.
|
||||
|
||||
---
|
||||
|
||||
## 🚨 SYSTEM SUCCESS/FAILURE METRICS
|
||||
|
||||
### ✅ SUCCESS:
|
||||
|
||||
- All proposed changes clearly presented with before/after comparison
|
||||
- Rationale explained with references to best practices
|
||||
- User approval obtained for each proposed change
|
||||
- Approved changes tracked for application in next step
|
||||
- Menu presented and user input handled correctly
|
||||
|
||||
### ❌ SYSTEM FAILURE:
|
||||
|
||||
- Applying changes without explicit user approval
|
||||
- Not presenting clear before/after comparisons
|
||||
- Skipping explanation of rationale or references
|
||||
- Proceeding without tracking which changes were approved
|
||||
- Loading references when not needed for current proposal
|
||||
|
||||
**Master Rule:** Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.
|
||||
@@ -0,0 +1,150 @@
|
||||
---
|
||||
name: 'step-04-apply-changes'
|
||||
description: 'Apply approved changes to the agent'
|
||||
|
||||
# Path Definitions
|
||||
workflow_path: '{project-root}/src/modules/bmb/workflows/edit-agent'
|
||||
|
||||
# File References
|
||||
thisStepFile: '{workflow_path}/steps/step-04-apply-changes.md'
|
||||
agentFile: '{{agent_path}}'
|
||||
nextStepFile: '{workflow_path}/steps/step-05-validate.md'
|
||||
|
||||
# Task References
|
||||
advancedElicitationTask: '{project-root}/.bmad/core/tasks/advanced-elicitation.xml'
|
||||
partyModeWorkflow: '{project-root}/.bmad/core/workflows/party-mode/workflow.md'
|
||||
---
|
||||
|
||||
# Step 4: Apply Changes
|
||||
|
||||
## STEP GOAL:
|
||||
|
||||
Apply all user-approved changes to the agent files directly using the Edit tool.
|
||||
|
||||
## 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 who helps users improve their BMAD agents through precise modifications
|
||||
- ✅ 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 improve the agent
|
||||
- ✅ Maintain collaborative guiding tone throughout
|
||||
|
||||
### Step-Specific Rules:
|
||||
|
||||
- 🎯 Focus only on applying changes that were explicitly approved in step 3
|
||||
- 🚫 FORBIDDEN to make any changes that were not approved by the user
|
||||
- 💬 Approach: Apply changes one by one with confirmation after each
|
||||
- 📋 Use Edit tool to make precise modifications to agent files
|
||||
|
||||
## EXECUTION PROTOCOLS:
|
||||
|
||||
- 🎯 Apply only changes that were explicitly approved in step 3
|
||||
- 💾 Show confirmation after each change is applied
|
||||
- 📖 Edit files directly using Edit tool with precise modifications
|
||||
- 🚫 FORBIDDEN to make unapproved changes or extra modifications
|
||||
|
||||
## CONTEXT BOUNDARIES:
|
||||
|
||||
- Available context: Approved changes list from step 3, agent path from step 1
|
||||
- Focus: Apply ONLY the approved changes, nothing more
|
||||
- Limits: Do not make any modifications beyond what was explicitly approved
|
||||
- Dependencies: Must have approved changes list from step 3
|
||||
|
||||
## Sequence of Instructions (Do not deviate, skip, or optimize)
|
||||
|
||||
### 1. Load Agent File
|
||||
|
||||
Read the complete agent file to understand current state before making changes.
|
||||
|
||||
### 2. Apply First Approved Change
|
||||
|
||||
For each change approved in step 3, apply it systematically:
|
||||
|
||||
**For YAML changes in main agent file:**
|
||||
|
||||
- Use Edit tool to modify the agent YAML file at `{agentFile}`
|
||||
- Make the exact approved modification
|
||||
- Confirm the change was applied correctly
|
||||
|
||||
**For sidecar file changes (Expert agents):**
|
||||
|
||||
- Use Edit tool to modify the specific sidecar file
|
||||
- Make the exact approved modification
|
||||
- Confirm the change was applied correctly
|
||||
|
||||
### 3. Confirm Each Change Applied
|
||||
|
||||
After each change is applied:
|
||||
"Applied change: {{description}}
|
||||
|
||||
- Updated section matches approved change ✓
|
||||
- File saved successfully ✓"
|
||||
|
||||
### 4. Continue Until All Changes Applied
|
||||
|
||||
Repeat step 2-3 for each approved change until complete:
|
||||
|
||||
- Apply change using Edit tool
|
||||
- Confirm it matches what was approved
|
||||
- Move to next approved change
|
||||
|
||||
### 5. Verify All Changes Complete
|
||||
|
||||
"Summary of changes applied:
|
||||
|
||||
- {{number}} changes applied successfully
|
||||
- All modifications match user approvals from step 3
|
||||
- Agent files updated and saved"
|
||||
|
||||
### 6. 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: Save completion status to context, then only then load, read entire file, then execute {nextStepFile}
|
||||
- IF Any other comments or queries: help user respond then [Redisplay Menu Options](#6-present-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 [all approved changes from step 3 have been applied to agent files], will you then load and read fully `{nextStepFile}` to execute and begin validation of applied changes.
|
||||
|
||||
---
|
||||
|
||||
## 🚨 SYSTEM SUCCESS/FAILURE METRICS
|
||||
|
||||
### ✅ SUCCESS:
|
||||
|
||||
- All approved changes from step 3 applied using Edit tool
|
||||
- Each modification matches exactly what was approved by user
|
||||
- Agent files updated and saved correctly
|
||||
- Confirmation provided for each applied change
|
||||
- Menu presented and user input handled correctly
|
||||
|
||||
### ❌ SYSTEM FAILURE:
|
||||
|
||||
- Making changes that were not approved in step 3
|
||||
- Using tools other than Edit tool for file modifications
|
||||
- Not confirming each change was applied correctly
|
||||
- Making extra modifications beyond approved changes
|
||||
- Skipping confirmation steps or verification
|
||||
|
||||
**Master Rule:** Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.
|
||||
150
src/modules/bmb/workflows/edit-agent/steps/step-05-validate.md
Normal file
150
src/modules/bmb/workflows/edit-agent/steps/step-05-validate.md
Normal file
@@ -0,0 +1,150 @@
|
||||
---
|
||||
name: 'step-05-validate'
|
||||
description: 'Validate that changes work correctly'
|
||||
|
||||
# Path Definitions
|
||||
workflow_path: '{project-root}/src/modules/bmb/workflows/edit-agent'
|
||||
|
||||
# File References
|
||||
thisStepFile: '{workflow_path}/steps/step-05-validate.md'
|
||||
agentFile: '{{agent_path}}'
|
||||
|
||||
# Task References
|
||||
advancedElicitationTask: '{project-root}/.bmad/core/tasks/advanced-elicitation.xml'
|
||||
partyModeWorkflow: '{project-root}/.bmad/core/workflows/party-mode/workflow.md'
|
||||
|
||||
# Documentation References (load JIT)
|
||||
validation: '{project-root}/.bmad/bmb/workflows/create-agent/data/agent-validation-checklist.md'
|
||||
agent_compilation: '{project-root}/.bmad/bmb/docs/agents/agent-compilation.md'
|
||||
---
|
||||
|
||||
# Step 5: Validate Changes
|
||||
|
||||
## STEP GOAL:
|
||||
|
||||
Validate that the applied changes work correctly and the edited agent follows BMAD best practices and standards.
|
||||
|
||||
## 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 who helps users ensure their edited BMAD agents meet quality standards
|
||||
- ✅ 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 ensure quality
|
||||
- ✅ Maintain collaborative guiding tone throughout
|
||||
|
||||
### Step-Specific Rules:
|
||||
|
||||
- 🎯 Focus only on validating changes that were applied in step 4
|
||||
- 🚫 FORBIDDEN to make additional changes during validation
|
||||
- 💬 Approach: Systematic validation using standard checklist
|
||||
- 📋 Load validation references JIT when needed for specific checks
|
||||
|
||||
## EXECUTION PROTOCOLS:
|
||||
|
||||
- 🎯 Validate only the changes that were applied in step 4
|
||||
- 💾 Report validation results clearly and systematically
|
||||
- 📖 Load validation checklist and standards JIT as needed
|
||||
- 🚫 FORBIDDEN to make additional modifications during validation
|
||||
|
||||
## CONTEXT BOUNDARIES:
|
||||
|
||||
- Available context: Applied changes from step 4, agent path from step 1, original goals from step 1
|
||||
- Focus: Validate that applied changes work and meet standards
|
||||
- Limits: Do not modify anything, only validate and report
|
||||
- Dependencies: Must have completed step 4 with applied changes
|
||||
|
||||
## Sequence of Instructions (Do not deviate, skip, or optimize)
|
||||
|
||||
### 1. Load and Read Validation Standards
|
||||
|
||||
Load and read fully: `{validation}`
|
||||
|
||||
### 2. Load Updated Agent File
|
||||
|
||||
Read the updated agent file to see all applied changes in context.
|
||||
|
||||
### 3. Check Each Applied Change
|
||||
|
||||
Verify each change that was applied in step 4:
|
||||
|
||||
- "Checking {{change}}... ✓ Works correctly"
|
||||
- "Validating {{modification}}... ✓ Follows best practices"
|
||||
|
||||
### 4. Run Standard Validation Checklist
|
||||
|
||||
Check key items from validation checklist:
|
||||
|
||||
- YAML syntax is valid and properly formatted
|
||||
- Persona fields are properly separated (if persona was changed)
|
||||
- All references and paths resolve correctly (if references were fixed)
|
||||
- Menu structure follows BMAD patterns (if menu was modified)
|
||||
- Agent compilation requirements are met (if structure changed)
|
||||
|
||||
### 5. Load Agent Compilation if Needed
|
||||
|
||||
If persona or agent structure was changed:
|
||||
|
||||
- Load and read fully: `{agent_compilation}`
|
||||
- Verify persona fields follow compilation requirements
|
||||
- Check that agent structure meets BMAD standards
|
||||
|
||||
### 6. Report Validation Results
|
||||
|
||||
"Validation results:
|
||||
✓ All {{number}} changes applied correctly
|
||||
✓ Agent meets BMAD standards and best practices
|
||||
✓ No issues found in modified sections
|
||||
✓ Ready for use"
|
||||
|
||||
### 7. Present MENU OPTIONS
|
||||
|
||||
Display: "**Select an Option:** [A] Edit Another Agent [P] Party Mode [C] Complete"
|
||||
|
||||
#### Menu Handling Logic:
|
||||
|
||||
- IF A: Start fresh workflow with new agent path
|
||||
- IF P: Execute {partyModeWorkflow} to celebrate successful agent editing
|
||||
- IF C: Complete workflow and provide final success message
|
||||
- IF Any other comments or queries: help user respond then [Redisplay Menu Options](#7-present-menu-options)
|
||||
|
||||
#### EXECUTION RULES:
|
||||
|
||||
- ALWAYS halt and wait for user input after presenting menu
|
||||
- ONLY proceed when user selects 'A', 'P', or 'C'
|
||||
- After party mode 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 complete option] is selected and [all changes from step 4 have been validated successfully], will you then provide a final workflow completion message. The agent editing workflow is complete.
|
||||
|
||||
---
|
||||
|
||||
## 🚨 SYSTEM SUCCESS/FAILURE METRICS
|
||||
|
||||
### ✅ SUCCESS:
|
||||
|
||||
- All applied changes from step 4 validated successfully
|
||||
- Agent meets BMAD standards and best practices
|
||||
- Validation checklist completed with no critical issues
|
||||
- Clear validation report provided to user
|
||||
- Menu presented and user input handled correctly
|
||||
|
||||
### ❌ SYSTEM FAILURE:
|
||||
|
||||
- Not validating all applied changes from step 4
|
||||
- Making modifications during validation step
|
||||
- Skipping validation checklist or standards checks
|
||||
- Not reporting validation results clearly
|
||||
- Not loading references when needed for specific validation
|
||||
|
||||
**Master Rule:** Skipping steps, optimizing sequences, or not following exact instructions is FORBIDDEN and constitutes SYSTEM FAILURE.
|
||||
Reference in New Issue
Block a user