git/AutoMetabuilder
1
0
Fork 0
mirror of https://github.com/johndoe6345789/AutoMetabuilder.git synced 2026-04-24 13:54:59 +00:00
Code Issues Packages Projects Releases Wiki Activity
Files
de876ab1301d40255918c1f1aa267ad3c300ed3d
AutoMetabuilder/ROADMAP.md
johndoe6345789 de876ab130 Introduce AutoMetabuilder core components and workflow packages: 
- Implement core components: CLI argument parsing, environment loading, GitHub service creation, and logging configuration.
- Add support for OpenAI client setup and model resolution.
- Develop SDLC context loader from GitHub and repository files.
- Implement workflow context and engine builders.
- Introduce major workflow packages: `game_tick_loop` and `contextual_iterative_loop`.
- Update localization files with new package descriptions and labels.
- Streamline web navigation by loading items from a dedicated JSON file.
2026-01-10 01:30:29 +00:00

4.9 KiB
Raw Blame History

Roadmap

Phase 1: Foundation

  • Basic GitHub Integration (fetching issues/PRs)
  • Local YAML prompt loading
  • Tool-based SDLC operations (branch/PR creation)
  • Multi-language support for messages

Phase 2: Enhanced Context & Reasoning

  • Roadmap Awareness: Bot should explicitly read and update ROADMAP.md.
  • Repository Indexing: Implement a way to index the codebase for better context.
  • Declarative Task Processing: Move more logic into JSON/YAML specifications.
  • Feedback Loop: Support for the AI to read comments on PRs it created.

Phase 3: Advanced Automation (MVP)

  • Automated Testing: Integration with test runners to verify changes before PR.
  • Linting Integration: Automatically run and fix linting issues.
  • Multi-Model Support: Easily switch between different LLM providers.
  • CI/CD Integration: Github Actions to run AutoMetabuilder on schedule or trigger.

Phase 4: Optimization & Scalability

  • Dockerization: Provide a Dockerfile and docker-compose for easy environment setup. Added run_docker_task tool.
  • Extended Toolset: Add tools for dependency management (poetry) and file manipulation (read/write/edit).
  • Self-Improvement: Allow the bot to suggest and apply changes to its own prompt.yml or tools.json.
  • Robust Error Handling: Implement exponential backoff for API calls and better error recovery.
  • Monitoring & Logging: Structured logging and status reporting for long-running tasks.

Phase 5: Ecosystem & User Experience

  • Web UI: A simple dashboard to monitor tasks and approve tool executions. Enhanced with settings and translation management.
  • Plugin System: Allow users to add custom tools via a plugin directory.
  • Slack/Discord Integration: Command and notify the bot from chat platforms.

Phase 6: Advanced Web UI & Remote Control

  • Remote Command Execution: Trigger bot runs from the Web UI.
  • User Authentication: Secure the Web UI with login.
  • Visual Task Progress: Real-time progress bars for long-running tasks.

Phase 7: Workflow UX & Component Library

  • Node-Based Workflow Engine: Replace task steps with micro-plugin nodes (inputs/outputs, loops).
  • Workflow Templates: Package reusable workflow presets (blank, single pass, iterative loop, plan/execute/summarize).
  • Workflow Template Picker: AJAX-loaded catalog with localized labels/descriptions.
  • Atomic Jinja Components: Split dashboard/prompt/settings/translations/sidebar into single-macro files.
  • AJAX Navigation Data: Render sidebar links from API payload with a client-side fallback.
  • Node Palette + Search: Categorized plugin library with search and click-to-add.

Phase 8: Modern Frontend Platform

  • Flask + Next.js split: Replace the Jinja-based FastAPI UI with a Flask REST backend and Next.js frontend consuming metadata, translations, workflows, logs, and nav via AJAX.
  • Atomic Next sections: Compose dashboard, workflow builder, prompt editor, settings, and translation editor into dedicated components powered by localized strings.
  • Workflow templates & navigation JSON: Serve workflow packages, nav items, and translation mappings from metadata-backed JSON endpoints.
  • Document build constraints: Record that next build --webpack fails in this sandbox because bundlers attempt to bind new ports, and continue iterating locally.
  • Storybook + Playwright: Add a Storybook catalog for the Material UI sections and a Playwright suite (with npm run test:e2e) so the frontend gets visual regression/backstop coverage tied to the Flask API.
  • Material UI + webhooks: Drive the dashboard with Material UI surfaces and a lightweight webhook emitter/listener so downstream components can react to run events without prop drilling.

Phase 9: Visual Workflow Canvas

  • n8n-Style Visual Workflow Canvas (Breakdown): Capture node + edge details so the canvas understands the micro-plugin graphs.
  • Canvas Layout Engine: DAG layout, zoom/pan, and background grid to keep large graphs navigable.
  • Palette Tags + Drag-to-Canvas: Tag nodes, add drag handles, and allow drag/drop placement onto the canvas.
  • Atomic Node Cards: Compact tile UI with status badges, icons, and inline rename/edit actions.
  • Ports + Connectors: Visualize input/output ports with validation and JSON metadata.
  • Edge Routing + Mini Map: Orthogonal edges with hover/selection states plus a mini overview map.
  • Selection + Inspector: Multi-select, bulk edit, right-side inspector for node/edge properties.
  • Inline Validation & Execution Preview: Warn on missing inputs and simulate data flow + store bindings.
  • Workspace Controls: Auto-save drafts, template import/export, keyboard shortcuts, undo/redo stack, context menus, and performant rendering for big graphs.
Reference in New Issue View Git Blame Copy Permalink