How chaudharyviv/agile_crew_netapp works
1. Repository purpose
- This is a Streamlit-based AI planning app for NetApp storage initiatives.
- It turns a user-entered initiative description into a structured set of planning artifacts for migration/modernization work.
2. Primary runtime shape
- The app runs as a single Python/Streamlit frontend process.
- There is no separate backend service; the UI directly invokes the orchestration layer in src/file_storage_crew.
3. Top-level UI boundary
- streamlit_app.py is the entry point and owns the interactive user experience.
- It collects initiative inputs, triggers planning, displays progress/results, and exposes export/publish actions.
4. Orchestration boundary
- src/file_storage_crew/crew.py is the main workflow engine.
- It coordinates the sequential multi-agent plan generation described in the README: business brief → backlog → sprint plan → technical design → architecture review → risk assessment → automation strategy → executive summary.
5. Agent and task definition boundary
- src/file_storage_crew/config.py holds the agent/task configuration.
- This is where the repository likely centralizes the seven agent personas, their roles, and the task prompts that shape each generated document.
6. Integration boundary for Confluence
- src/file_storage_crew/confluence.py isolates Atlassian Confluence publishing.
- This keeps external API interaction separate from the planning workflow, so generation can run independently of publishing.
7. Data flow through the system
- User inputs enter through the Streamlit UI.
- The UI passes the initiative context into the CrewAI workflow.
- Each agent produces an intermediate artifact that becomes input to the next step.
- The final outputs are collected as markdown documents and surfaced in the UI for download or publishing.
8. Generated artifact model
- The system produces eight named markdown deliverables, each representing a distinct planning layer.
- These artifacts form the primary output boundary of the application and are the main units for review, export, and Confluence page creation.
9. Secrets and configuration boundary
- .env.example, Streamlit secrets, and pydantic-settings/dotenv usage indicate configuration is externalized.
- OPENAI_API_KEY is mandatory; Confluence credentials are optional and only activate the publishing path.
10. External services
- OpenAI is the core AI dependency behind CrewAI agent execution.
- Atlassian Confluence is an optional downstream system used to publish the generated plan hierarchy.
- Streamlit provides the web UI runtime rather than a traditional REST API.
11. Packaging and local setup
- requirements.txt defines the Python dependency set.
- .devcontainer/ indicates the repo supports containerized development, which is useful for consistent local setup but does not change the application architecture.
12. Deployment shape
- The repo is structured for local execution and Streamlit Cloud deployment.
- Deployment.md and the Streamlit secrets model suggest the app is expected to run as a hosted single-page planning tool, not as a multi-service distributed system.
Main entry points: streamlit_app.py Streamlit app · crew.py Crew engine · confluence.py Confluence · .env.example Secrets
Components 18
UI
Interactive Streamlit entrypoint and user-facing workflow
- Streamlit app · Collects initiative inputs, runs planning, and presents results
streamlit_app.py - Secrets · Environment and Streamlit secret settings for local and hosted execution
.env.example
Planning Engine
CrewAI orchestration and agent/task definitions
- Crew engine · Coordinates the sequential multi-agent planning pipeline
src/file_storage_crew/crew.py - Agent config · Holds personas, task prompts, and workflow configuration
src/file_storage_crew/config.py - Initiative input · User-provided scope and constraints passed from the UI into the workflow
Integrations
External services and publishing boundary
- Confluence · Optional publishing path for generated artifacts
src/file_storage_crew/confluence.py - OpenAI · Model provider used by CrewAI agents to generate content
- Streamlit Cloud · Hosted runtime for the single-page planning app
.streamlit/config.toml - Dev container · Containerized local development environment
.devcontainer/devcontainer.json
Deliverables
Generated planning artifacts and outputs
- Business brief · First planning layer establishing business intent and context
- Backlog · Structured work items derived from the initiative brief
- Sprint plan · Sequenced implementation plan for delivery execution
- Technical design · Solution design and implementation details
- Architecture review · Checks design fit, boundaries, and target architecture
- Risk assessment · Identifies delivery, migration, and operational risks
- Automation strategy · Defines automation approach for migration and modernization work
- Exec summary · Concise leadership-facing recap of the plan
- Plan bundle · The complete markdown deliverables returned by the workflow
Connections 29
- Streamlit app → Initiative input: collects · Captures initiative description and planning inputs
- Streamlit app → Crew engine: starts · Launches the planning workflow from the UI
- Initiative input → Crew engine: feeds · Provides scoped initiative context to orchestration
- Crew engine → Agent config: uses · Loads personas, prompts, and task definitions
- Crew engine → OpenAI: calls · Executes agent reasoning and content generation
- Crew engine → Business brief: produces · Generates the first planning artifact
- Business brief → Backlog: refines into · Converts brief into structured work items
- Backlog → Sprint plan: sequences into · Turns backlog into execution planning
- Sprint plan → Technical design: informs · Shapes the solution design and implementation detail
- Technical design → Architecture review: reviewed by · Validates the design against architecture expectations
- Architecture review → Risk assessment: surfaces · Feeds review findings into risk analysis
- Risk assessment → Automation strategy: guides · Influences automation approach and priorities
- Automation strategy → Exec summary: summarized in · Rolls the plan into an executive-facing summary
- Exec summary → Plan bundle: bundles · Completes the final set of markdown outputs
- Business brief → Plan bundle: included in · Adds generated document to the final output set
- Backlog → Plan bundle: included in · Adds generated document to the final output set
- Sprint plan → Plan bundle: included in · Adds generated document to the final output set
- Technical design → Plan bundle: included in · Adds generated document to the final output set
- Architecture review → Plan bundle: included in · Adds generated document to the final output set
- Risk assessment → Plan bundle: included in · Adds generated document to the final output set
- Automation strategy → Plan bundle: included in · Adds generated document to the final output set
- Exec summary → Plan bundle: included in · Adds generated document to the final output set
- Streamlit app → Plan bundle: displays · Shows results and supports download/export
- Streamlit app → Confluence: publishes via · Optionally sends the generated hierarchy to Confluence
- Confluence → Plan bundle: uploads · Creates pages from the final markdown artifacts
- Secrets → OpenAI: configures · Provides API keys and runtime settings
- Secrets → Confluence: enables · Supplies optional Confluence credentials
- Dev container → Streamlit app: supports · Provides a reproducible local development environment
- Streamlit Cloud → Streamlit app: hosts · Deployment target for the single-process app