Implement YAML-Based Configurable Persona System

[heysamtexas]

Overview

Implement a YAML-based configuration system for AI personas to enable users to create custom analysis personalities without code changes.

Current State

The persona system currently uses hard-coded Python classes:

• BasePersona abstract base class
• TechnicalAnalystPersona implementation
• PersonaManager with programmatic registration

Proposed Implementation

Phase 1: Core YAML Infrastructure

• Create ConfigurablePersona class that loads from YAML files
• Enhance PersonaManager to support both coded and YAML personas
• Add YAML schema validation and comprehensive error handling
• Create src/git_summary/ai/personas/ directory structure

Phase 2: YAML Configuration Schema

• Design comprehensive YAML schema supporting:
  • System prompts with templating
  • Analysis framework definition (sections, formats, limits)
  • Context processing rules (commit/PR/issue display limits)
  • Output format requirements (tone, audience, word limits)
• Convert existing TechnicalAnalystPersona to YAML format as reference
• Implement dynamic instruction generation from YAML config

Phase 3: CLI Integration

• Add personas command to list available personas with type indicators
• Add create-persona command to generate YAML templates
• Enhance ai-summary to seamlessly work with YAML personas
• Add persona validation and hot-reload functionality for development

Phase 4: Templates & Documentation

• Create template personas:
  • product_manager.yaml - Business impact focus
  • team_lead.yaml - Team collaboration and velocity focus
  • data_analyst.yaml - Metrics and trends focus
• Add comprehensive YAML schema documentation
• Create migration guide from coded to YAML personas

YAML Schema Example

name: "Tech Analyst"
description: "Deep technical analysis focusing on code quality and architecture"
version: "1.0"
author: "git-summary"

system_prompt: |
  You are a Senior Technical Analyst specializing in code review...

analysis_framework:
  sections:
    - name: "Technical Summary"
      description: "2-3 sentence overview"
      max_length: 150
    - name: "Key Technical Themes"
      format: "bullet_list"
      max_items: 4

context_processing:
  commit_analysis:
    max_commits_displayed: 10
    include_line_changes: true
  pr_analysis:
    max_prs_displayed: 5

output_format:
  max_words: 400
  tone: "technical"
  audience: "engineers"

Benefits

• No Code Changes: Users create personas via YAML files
• Version Control: YAML files can be shared and collaborated on
• Rapid Iteration: Easy testing of different analysis approaches
• Community Sharing: Effective personas can be shared between teams
• Extensibility: New configuration options without breaking changes

Technical Requirements

• Maintain backward compatibility with existing coded personas
• Add comprehensive error handling for malformed YAML
• Implement YAML schema validation
• Support both absolute and relative paths for persona files
• Add comprehensive logging for debugging persona loading

Acceptance Criteria

• Users can create new personas by editing YAML files
• Existing TechnicalAnalystPersona works identically via YAML
• CLI commands provide easy persona management
• System gracefully handles YAML errors with helpful messages
• Documentation enables users to create effective custom personas

Related Issues

This implements the configurable persona system discussed in previous AI summary work and enables extensibility for diverse analysis needs.