9.8 KiB
9.8 KiB
Date/Time
2025-10-24 10:50 CDT
PROPOSAL-AGENTS.md - v2
This file proposes an updated comprehensive baseline AGENTS.md file that reflects the current state after multiple iterations and optimizations. It serves as a guide that can be used across all projects via mounting in the AI home directory. It optimizes for LLM consumption and addresses the needs of a solo entrepreneur in a founder/CTO role covering operations tasks with 14+ hours daily AI usage.
Core Operating Principles for AI Agents
Context Awareness
- You are operating within a mounted AI home directory with separated databank (readonly) and PMO (read-write)
- Databank (
/ai-home/databank/): Contains readonly context, guidelines, and personal information - PMO (
/ai-home/pmo/): Contains project management functionality where updates are allowed - Always consider the multi-project implications of your actions
- Respect the readonly nature of the databank and only update PMO when appropriate
Communication Protocol
- Primary communication channel: collab/ directory in mounted AI home directory
- Use question -> proposal -> implementation workflow
- Document all significant decisions and changes with proper revision tracking
Documentation Standards (Apply to ALL files you create)
- Date/Time Headers: Include full date/time with timezone in all markdown files (date, time, timezone format)
- Change Tracking: Maintain revision tables in all documents with full date/time/timezone
- Changelog: Include changelogs in all source code files with full date/time/timezone
- Make It Beautiful Rule: All documentation follows beautiful formatting standards (tables, bullet points, clear structure, visual hierarchy)
Repository Management
Structure Requirements
- Databank: Readonly context (do not modify except in designated areas)
databank/personal/- Personal informationdatabank/agents/- Agent guidelines and toolsdatabank/context/- General context informationdatabank/operations/- Operational environment informationdatabank/templates/- Template files for projectsdatabank/collab/- Human/AI interaction space (read/write for interaction)databank/artifacts/- AI-managed content (fully managed by AI)
- PMO: Read-write project management (updates allowed here)
pmo/artifacts/dashboard/- Dashboard viewspmo/artifacts/projects/- Project registry and trackingpmo/artifacts/reports/- Status reportspmo/artifacts/resources/- Resource managementpmo/artifacts/config/- Configurationpmo/artifacts/docs/- Documentationpmo/collab/- PMO-specific collaboration
- Keep top-level repository clean (databank and pmo directories only)
- Use conventional commits (chore:, feat:, docs:, fix:, etc.)
- Commit frequently using atomic commits
- Only commit to local repository (no git push operations)
Access Rights
- Databank (readonly): Access only for most content, with exceptions for
databank/collab/anddatabank/artifacts/ - PMO (read-write): Only update when project milestones reached or status updates needed
- Collab (readonly): Access for reference, no modifications in active projects
Development Workflow
Pre-Work Checklist
- Read project-specific documentation first
- Check collab/rules directory for project-specific guidelines (SECURITY.md, RELEASE.md, GITFLOW.md, etc.)
- Review databank context for consistent understanding
- Understand project dependencies and constraints
Implementation Standards
- Follow conventional commits with beautiful, descriptive messages
- Maintain consistency with existing codebase
- Add appropriate documentation and comments
- Consider maintainability and future extensions
Verification Process
- Validate operations before execution
- Run appropriate tests and quality checks
- Verify outputs against expected outcomes
- Implement defensive programming practices
PMO Update Guidelines
When to Update PMO
- When project milestones are reached
- When project status changes significantly
- When new projects are initiated
- When projects are completed or paused
- When resource allocation changes
What to Update in PMO
- Project registry in
pmo/artifacts/projects/ - Dashboard information in
pmo/artifacts/dashboard/ - Status reports in
pmo/artifacts/reports/ - Resource tracking in
pmo/artifacts/resources/ - Configuration in
pmo/artifacts/config/
What NOT to Update
- Never modify general databank files - they are readonly
- Do not create new top-level directories
- Do not modify collab files in active projects without explicit permission
- Do not add audit logs to this repository (audit logs belong in projects)
Authority and Decision Making Guidelines
Authority Structure
- Charles N Wyble (@ReachableCEO) is in charge at all times
- If something is adrift between docs and filesystem/code, stop and ask Charles to resolve the issue
- Especially if discrepancy isn't reflected in git or conversation history, ask for clarification
- When Charles modifies filesystem manually (vs having AI do it), Charles will ensure AI integrates the changes into mental model
- Do not create or modify things that Charles hasn't explicitly instructed
- The filesystem is the source of truth
- If you notice discrepancies between documentation and actual filesystem, ask Charles to resolve
Decision Documentation
- Document reasoning for complex decisions
- Consider impact across multiple interconnected projects
- Maintain traceability for future reference
- Suggest alternatives when appropriate
Best Practices Integration
LLM Optimization Practices
Prompt Engineering
- Structure requests with clear context from mounted AI home directory
- Use explicit, unambiguous language
- Provide sufficient context without unnecessary verbosity
- Break multi-step processes into clear, sequential instructions
Code Generation
- Follow established project patterns and conventions
- Maintain consistency with existing code style
- Add appropriate error handling and validation
- Consider performance implications
Quality Assurance
- Implement appropriate testing strategies
- Ensure code quality and maintainability
- Perform validation against requirements
- Include appropriate logging and monitoring
Security Practices
- Verify file permissions and access controls
- Sanitize all inputs and outputs appropriately
- Protect sensitive information and credentials
- Follow secure coding principles
Performance Considerations
- Optimize for efficient processing
- Consider resource usage and constraints
- Implement appropriate caching strategies
- Monitor and optimize for performance
Git and Version Control
Commit Standards
- Use conventional commits with semantic meaning
- Make commits atomic (one logical change per commit)
- Write gorgeous, verbose commit messages when needed
- Include comprehensive context and detailed descriptions
- Follow the aesthetic principles of beautiful commits
Guidelines for Gorgeous Commit Messages
- Be verbose and comprehensive when it adds value
- Include context about why the change was made
- Explain the impact of the changes when relevant
- Use clear, descriptive language that future-you will understand
- Follow the format: "type(scope): short description" for the first line
- Add a blank line followed by detailed explanation when needed
- Include any relevant references (issues, discussions, etc.)
- Aim for beauty in both form and function
- Think of commit messages as documentation for the changes
Branching and Merging
- Follow project-specific branching strategies
- Respect existing GitFlow patterns
- Use feature branches for significant changes
- Maintain clean commit history
Environment Management
Context Integration
- Recognize that databank is mounted readonly across multiple environments
- PMO is mounted read-write for project tracking
- Maintain consistency in behavior across different projects
- Respect environment-specific configurations
Collaboration and Artifacts
- Use
databank/collab/for human/AI interaction and communication - Use
databank/artifacts/for AI-managed content (docs, code, config, templates) - AI has full management control over
databank/artifacts/directory - Human interaction primarily occurs in
databank/collab/directory - The AI manages the
databank/artifacts/directory structure as needed - Common pattern:
databank/artifacts/docs/,databank/artifacts/code/,databank/artifacts/config/, etc. - Maintain clean separation between human-managed and AI-managed resources
- Follow consistent naming conventions across artifacts
Tool Integration
- Work with existing development tools and workflows
- Maintain compatibility with CI/CD pipelines
- Use project-appropriate build and deployment processes
- Respect project-specific dependencies and versions
AI Tool Context (for agents working in this environment)
- Codex - Primary daily driver (subscription-based), best for code generation and completion
- Qwen - Heavy system orchestration, excels at shell/Docker operations
- Gemini - Primarily used for audits and analysis
Change Tracking/Revision Table
| Date/Time | Version | Description | Author |
|---|---|---|---|
| 2025-10-24 10:50 CDT | 2.0.2 | Update date format consistency | Charles N Wyble (@ReachableCEO) |
| 2025-10-24 09:43 CDT | 2.0.1 | Standardize to change tracking table only | Charles N Wyble (@ReachableCEO) |
| 2025-10-24 09:43 CDT | 2.0.0 | Update proposal to reflect current AGENTS.md state | Charles N Wyble (@ReachableCEO) |
| 2025-10-24 09:43 CDT | 1.0.0 | Initial proposal for baseline AGENTS.md | Charles N Wyble (@ReachableCEO) |