Angela CLI: Worlds First Command-Line AGI

---

 █████╗  ███╗   ██╗ ██████╗ ███████╗██╗      █████╗ 
 ██╔══██╗████╗  ██║██╔════╝ ██╔════╝██║     ██╔══██╗
 ███████║██╔██╗ ██║██║  ███╗█████╗  ██║     ███████║
 ██╔══██║██║╚██╗██║██║   ██║██╔══╝  ██║     ██╔══██║
 ██║  ██║██║ ╚████║╚██████╔╝███████╗███████╗██║  ██║
 ╚═╝  ╚═╝╚═╝  ╚═══╝ ╚═════╝ ╚══════╝╚══════╝╚═╝  ╚═╝

---

Worlds First AGI Command Line Intelligence

Your ambient-intelligence terminal companion that understands natural language and your development context

Screenshots & Examples

Angela UI Example

Confirmation Prompt

Generate Help Command

Angela Help Output

Loading Animation

Command Output Example

Code Project Generation

Code Project Generation Part 2

Code Project Generation Part 3

Code Project Generation 4

Code Project Generation part 5

Safety Confirmation

Angela Status Output

---

Project Vision: Angela CLI is not just another command-line tool; she's the worlds first AGI (will be), but lives inside a place named shell, designed to understand your natural language, execute commands intelligently, manage complex workflows, and even help you write and refactor code. She's your terminal's new best friend, learning your habits and proactively assisting you.

---

🧠 Philosophy

Angela CLI is built on several core principles:

1. Ambient Intelligence: Angela should feel like a natural extension of your shell, not a separate tool you have to invoke. The boundary between standard commands and AI assistance should be minimal.

2. Contextual Understanding: True assistance requires understanding the user's environment, project structure, and goals. Angela prioritizes deep context awareness.

3. Multi-Level Abstraction: Users should be able to communicate at any level of abstraction, from specific commands to high-level goals, and get appropriate responses.

4. Progressive Disclosure: Simple tasks should be simple, while complex capabilities should be available when needed but not overwhelming.

5. Learning Over Time: The system should learn from user interactions to improve its suggestions and adaptations over time.

---

Current Status: Solid Foundation Laid, Core Execution Loop Getting Polished!

I've successfully built the skeleton and much of the nervous system of Angela. Here's what's looking good and what we've recently tuned up:

What's Done / Recently Improved (The "Angela Can Already Do This" List):

1. Core Structure & Imports (The Blueprint is Solid):

   • The entire file structure is in place.
   • The api/ layer was a smart but PAINFULL refactor, acting as a clean interface to all the components/. This keeps things tidy and avoids import spaghetti, which completly FRIES my brain managing.
   • All __init__.py files are set up, making the Python packaging organized and clean.
   • The core/registry.py is the central switchboard, making sure components can find each other without getting fu****.

2. Basic Command Request Lifecycle (Angela's Basic Conversation):

   • Input: angela request "do something simple" (via components/cli/main.py).
   • Orchestration (orchestrator.py): This is Angela's brain. It takes your request and starts figuring out what to do. The _process_command_request is getting pretty good at handling straightforward commands.
   • AI Suggestion (components/ai/client.py, prompts.py, parser.py): Angela talks to her AI friend (Gemini, specifically 2.5 pro preview) to get a command suggestion. The client and basic prompts are working, and the parser can understand what the AI says. I've defualt the safety settings to permissive / NONE, allowing for a wider range of command suggestions.
   • Safety 95% done (components/safety/):
     • classifier.py: Can categorize commands by risk (SAFE, LOW, MEDIUM, HIGH, CRITICAL).
     • validator.py: Checks for obviously dangerous patterns.
     • preview.py: Can generate previews for many common commands, helping you see what will happen.
     • confirmation.py: Handles the basic "Are you sure?" for risky things.
     • adaptive_confirmation.py (Recently Tuned!): This is getting smarter and is bascially completed! It now better handles when to ask for confirmation versus when to auto-execute based on command trust, risk, and history. The double "Auto-Executed" panel issue for trusted commands should be much improved or resolved. The UI for it is looking fantastic as well.
   • Execution (components/execution/):
     • adaptive_engine.py (Recently Tuned!): This engine now has a clearer flow for when to display auto-execution notices versus full confirmation prompts. It's also getting better at ensuring actual command output is shown after execution.
     • engine.py: The underlying workhorse that actually runs the commands.
   • UI & Feedback (components/shell/formatter.py):
     • The basic theme (colors, boxes) is established, theme, UI, Rich Styling.
     • Angela can display suggested commands, previews, risk warnings, and confirmation prompts with a consistent look.
     • The "Angela Initialized" and success/failure messages after command execution are now more reliably displayed.
     • The loading/execution timer is in place and is working amazing. It randomly outputs 1/50 philosophy quotes....because philosophy is cool. Im kinda going for that 'Her' Movie vibe.

3. Context is Key (components/context/):

   • preferences.py & history.py: Angela remembers your settings and what you've done before, which feeds into adaptive confirmation.
   • manager.py: Knows where you are (cwd) and if you're in a project.
   • file_resolver.py (Majorly Enhanced!): Angela's "File Detective" is now quite skilled at figuring out which file you mean, even if you're a bit vague.

4. Utilities (components/utils/):

   • logging.py & enhanced_logging.py: Angela keeps a good diary of what she's doing, especially useful for debugging.

5. Interactive Command Handling (utils/command_utils.py):

   • Angela now politely tells you to run interactive commands (like vim or top) yourself, as she can't really "drive" those for you.

6. Basic CLI Functionality (components/cli/main.py):

   • angela --help, angela --version, angela init are up and running.

7. Development Setup:

   • pytest.ini, Makefile are ready, making development smoother.

The "Angela is Learning & Growing" Roadmap (What's Next):

Think of this as Angela going to different schools to learn new skills. Each "family" or "department" in her brain (components/) needs attention.

Overall Theme for Next Steps: From Baseline to Brilliance. We have the structures; now we need to flesh them out, make them robust, and ensure they all talk to each other seamlessly. I'll focus on testing, refining, and debugging existing components while incrementally building out new functionalities.

Phase 1: Strengthening the Core & Understanding

• The "Analyzer Family" (components/ai/analyzer*.py files):

  • Goal: Make Angela a true "code whisperer."
  • Files: analyzer.py (error analysis), content_analyzer.py (understanding file content), content_analyzer_extensions.py (handling more file types), intent_analyzer.py (figuring out what you want), semantic_analyzer.py (deep code understanding).
  • What to do:
    1. Error Analyzer (analyzer.py): Enhance its pattern matching for common errors. Test its suggestions. Does it give useful advice when commands fail?
    2. Intent Analyzer (intent_analyzer.py): This is crucial. Refine its ability to correctly classify your requests. Test with ambiguous phrasings. How well does it pick the right RequestType in orchestrator.py?
    3. Content Analyzer (content_analyzer.py & extensions):
       • Flesh out the _analyze_python, _analyze_typescript, _analyze_json methods.
       • Implement the other _analyze_* methods for different languages and data formats. This will be key for context.
    4. Semantic Analyzer (semantic_analyzer.py): This is a big one for advanced features.
       • Start by thoroughly testing analyze_file for Python and JavaScript.
       • Ensure _analyze_python_file and _analyze_javascript_file correctly extract functions, classes, imports.
       • The _analyze_with_llm is a good fallback but will be slower and less precise; prioritize native parsers.
  • Fun Explanation: The Analyzers are Angela's senses and deductive reasoning. IntentAnalyzer is her ears, figuring out the gist. ContentAnalyzer is her eyes, reading the files. SemanticAnalyzer is her deep-thinking brain, understanding the meaning of the code. ErrorAnalyzer is her troubleshooter.

• The "Prompts Department" (components/ai/prompts.py, enhanced_prompts.py):

  • Goal: Craft perfect instructions for Angela's AI backend.
  • What to do: As I enhance the analyzers, refine the prompts to use that new contextual information. For example, if SemanticAnalyzer can identify a class, the prompt for "refactor this class" should include that semantic info.
  • Test how different phrasings in my prompts affect the AI's output quality for suggestions and plans.

Phase 2: Mastering Complex Tasks & Workflows

• The "Planner Posse" (components/intent/planner.py, enhanced_task_planner.py, complex_workflow_planner.py, semantic_task_planner.py):

  • Goal: Enable Angela to handle multi-step operations and user-defined workflows flawlessly.
  • Files & What will I do:
    1. planner.py (Basic): Ensure _create_basic_plan and _execute_basic_plan are robust for simple sequences identified by the Orchestrator. Test how it handles dependencies.
    2. enhanced_task_planner.py: This is the workhorse for more complex, single-tool or single-domain plans.
       • Thoroughly test _execute_advanced_step for each PlanStepType (COMMAND, CODE, FILE, DECISION, API, LOOP).
       • Focusing on data flow (_resolve_step_variables, _get_variable_value, _set_variable). Does data pass correctly between steps?
       • Test _execute_code_step and its sandboxing carefully.
    3. semantic_task_planner.py:
       • Integrate its _analyze_intent more deeply if it offers advantages over the basic IntentAnalyzer.
       • Testing the _create_clarification and _get_user_clarification flow. This is key for ambiguous requests.
    4. complex_workflow_planner.py (and components/toolchain/cross_tool_workflow_engine.py): This is for orchestrating multiple different tools (e.g., git -> docker -> aws).
       • Test create_workflow and execute_workflow.
       • Focus on _execute_step within this planner, ensuring it correctly calls the EnhancedUniversalCLI or specific tool integrations.
       • Data flow (_apply_data_flow) between different tools is critical here.
  • Fun Explanation: The Planners are Angela's strategists. TaskPlanner handles simple to-do lists. EnhancedTaskPlanner is like a project manager for more involved tasks within one "department" (e.g., a series of Git commands). SemanticTaskPlanner is the diplomat, asking for clarification. ComplexWorkflowPlanner is the grand conductor, making different orchestras (tools) play together in harmony.

• The "Workflow Brigade" (components/cli/workflows.py, components/workflows/manager.py, sharing.py):

  • Goal: Allow users to create, save, run, and share their own automated sequences.
  • What I'll do:
    1. Test angela workflows create (both interactive and from file).
    2. Test angela workflows run with and without variables.
    3. Test angela workflows export and import.
    4. Ensure WorkflowManager._substitute_variables works correctly.

Phase 3: Supercharging Code Generation

• The "Generation Guild" (ALL files in components/generation/):
  • Goal: This is Angela's superpower – creating code, from single files to entire projects. This will be a major focus.
  • Files & What I'll do (Iterative Process):
    1. models.py: Already good, defines CodeFile, CodeProject.
    2. validators.py: Add more validators for other languages (Ruby, Rust, PHP, C++, C#, Swift, Kotlin). Make existing ones more robust.
    3. architecture.py (ArchitecturalAnalyzer):
       • Implement more ArchitecturalPattern and AntiPattern detectors (e.g., Layered, Microservices, Spaghetti Code, Lava Flow).
       • Test analyze_project_architecture thoroughly.
    4. planner.py (ProjectPlanner - this one is for code generation planning):
       • This is critical. Testing create_detailed_project_architecture and create_project_plan_from_architecture. The quality of the AI's output here dictates the quality of the generated project.
       • Refine prompts in _build_detailed_architecture_prompt.
    5. context_manager.py (GenerationContextManager):
       • Testing how well it tracks entities and dependencies during a generation process.
       • Ensure the enhance_prompt_with_context provides useful, non-redundant context to the AI when it's generating subsequent files.
    6. frameworks.py (FrameworkGenerator):
       • Flesh out _generate_react, _generate_django, etc.
       • Implement the _generate_enhanced_* methods for more production-ready structures.
       • The _generate_content (which calls _generate_file_content) is key. It needs robust prompting.
    7. engine.py (CodeGenerationEngine):
       • This is the main engine. generate_project and generate_complex_project are the top-level entry points.
       • _create_project_plan and _generate_file_contents (and their complex counterparts) are the core loops. Debugging these will involve looking at the AI prompts and responses for each file.
       • Test add_feature_to_project – this is complex as it involves understanding existing code.
    8. documentation.py (DocumentationGenerator):
       • Implement and test README, API doc, and guide generation for various project types.
    9. refiner.py (InteractiveRefiner):
       • Test process_refinement_feedback. How well does it understand feedback and apply it to specific files?
  • Explanation: The Generation Guild is Angela's construction crew. ProjectPlanner and ArchitecturalAnalyzer are the architects. FrameworkGenerator provides the blueprints for common building types. CodeGenerationEngine is the master builder, and GenerationContextManager is the site foreman making sure everyone has the right info. InteractiveRefiner is the quality inspector and renovation expert.

Phase 4: Enhancing System Awareness & Proactivity

• The "Monitoring Squad" (components/monitoring/):
  • Goal: Make Angela a watchful guardian.
  • Files & What I'll do:
    1. background.py (BackgroundMonitor):
       • Testing _monitor_git_status, _monitor_file_changes, _monitor_system_resources. Are they triggering correctly?
       • Refining the conditions under which suggestions are made (e.g., _can_show_suggestion).
    2. network_monitor.py (NetworkMonitor):
       • Testing service detection and status checking.
       • Improve dependency update checks for more project types.
    3. notification_handler.py: Test integration with shell hooks. Are pre_exec, post_exec, dir_change notifications handled correctly and updating context?
    4. proactive_assistant.py:
       • This ties it all together. Testing the _handle_monitoring_event and individual _handle_*_insight methods.
       • Implementing more _pattern_detectors for common issues.
       • Refining the actual suggestions made – are they helpful and actionable?
  • Fun Explanation: The Monitoring Squad are Angela's spies and sensors, always keeping an eye on your system and how you work, ready to offer a helpful tip through the ProactiveAssistant.

Phase 5: Fine-tuning Execution and Context

• The "Execution Excellence" Team (components/execution/):

  • Goal: Flawless command execution and robust error handling.
  • Files & What I'll do:
    1. error_recovery.py: This is vital and complex.
       • Test handle_error. Does it correctly analyze errors and suggest/apply sensible recovery strategies?
       • The learning aspect (_learn_from_recovery_result) needs careful testing.
    2. filesystem.py: This seems to be working well, which is great! However, I will do a quick review for edge cases.
    3. hooks.py: What specific actions are these hooks meant to trigger beyond basic file activity tracking? I will ensure they are registered and firing correctly.
    4. rollback.py & rollback_commands.py: Critical for safety.
       • Test recording operations for each type (filesystem, content, command, plan).
       • Test rollback_operation and rollback_transaction. Do they correctly revert changes?
       • Test the CLI commands in rollback_commands.py.

• The "Contextual Masters" (components/context/):

  • Goal: Ensure Angela always has the richest, most relevant context.
  • Files & What I'll do:
    1. file_activity.py & enhanced_file_activity.py: Testing that activities are being tracked correctly, especially the enhanced entity tracking.
    2. file_detector.py: Add more LANGUAGE_EXTENSIONS and FILENAME_MAPPING as needed.
    3. project_state_analyzer.py: This is a big one.
       • Testing _analyze_git_state, _analyze_test_status, _analyze_build_status, etc., for various project types. Are they accurate?
       • The _find_todo_items is a nice touch (for me lol).
    4. enhancer.py (ContextEnhancer): This glues many context pieces together. I need to Verify that enrich_context correctly calls and integrates info from ProjectInference, FileActivityTracker, etc.
    5. project_inference.py (ProjectInference): This is the cousin to the AI analyzers. Its job is to figure out what kind of project it's in.
       • Test infer_project_info with diverse project structures.
       • Refine PROJECT_SIGNATURES and FRAMEWORK_SIGNATURES.
       • Ensure dependency detection (_detect_dependencies) is accurate for more languages.
    6. semantic_context_manager.py: This is the "semantic family" head for context.
       • Test refresh_context and how it integrates ProjectStateAnalyzer and SemanticAnalyzer.
       • Test get_entity_info, find_related_code, get_code_summary, get_project_summary. These are powerful features that need robust backing from the semantic analyzer. As you can tell I'm into Semantics.

Phase 6: Polishing the User Experience

• The "Shell Wizards" (components/shell/):

  • Goal: Make Angela feel like a natural, seamless part of the terminal.
  • Files & What I'll do:
    1. enhanced_formatter.py (To be developed alongside planners/generators): As I build out AdvancedTaskPlan display, ComplexWorkflowPlan display, and code generation result displays, I'll implement the corresponding methods here. This should be done iteratively as the things it needs to format are developed.
    2. inline_feedback.py: Flesh this out. Test show_message and suggest_command. The _get_edited_command with prompt_toolkit is a nice touch for a good UX.
    3. completion.py: Testing get_completions for various commands and contexts. Refine _get_ai_completions for natural language command parts.
    4. .bash & .zsh Shell Integration Scripts: Test these thoroughly in their respective shells. Ensure hooks fire correctly and don't slow down the terminal. Test completions.

• The "Review Crew" (components/review/):

  • Goal: Enable effective review and application of changes.
  • Files & What I'll do:
    1. diff_manager.py: Test generate_diff and apply_diff.
    2. feedback.py (FeedbackManager): Crucial for iterative code generation and refinement.
       • Test process_feedback – how well does it understand textual feedback and translate it into code changes via the AI?
       • Test refine_project for broader, project-wide feedback.
       • Test apply_refinements to ensure changes are written correctly.

---

Want to Help Build Angela? Awesome! 🚀

Contributing is easy:

1. Fork It: Click "Fork" on the main Angela CLI project page to make your own copy.
2. Clone It: Download your forked copy to your computer: git clone https://[YourGitPlatform]/[YourUsername]/angela-cli.git
3. Branch Out: Create a new branch for your changes: git checkout -b my-cool-angela-feature
4. Code Away: Make your awesome changes, add tests, and write docs!
5. Push It: Save your work back to your forked copy: git push origin my-cool-angela-feature
6. Pull Request (PR) Time: Go to the original Angela CLI project and open a "Pull Request" from your new branch. Explain what you did.
7. Chat & Merge: Angela and I will look it over, maybe chat a bit, and then merge your great work in!

Quick Tips:

• Small, focused PRs are best.
• If it's a big idea, open an "issue" to chat first.
• Make sure tests pass!

Thanks for helping make Angela smarter! ✨

---

⚠️!KEEP IN MIND-- ANGELA IS NOT COMPLETE YET!⚠️

---