Business Workflow Map
Understand end-to-end business workflows and how your code implements each step.
Codebase analysis · Business workflows
Trace business workflows in complex codebases.
UnvibeCode reverse-engineers complex codebases into business workflows, connected code, edge cases, and evidence-backed risks.
Your LLM wrote the code. Do you know what it actually built?
pip install --upgrade unvibecodePython 3.11+ / Windows, macOS & Linux
SEE UNVIBECODE IN ACTION
One review. Four practical outputs.
Connected Code Map
Trace imports, symbols, and static code relationships. Hover to preview connected code; click a file to download a focused LLM context package.
Business Risk Findings
Review evidence-backed risks tied to analyzed workflows and supporting code.
Complete Repository Context
Download the complete normalized repository ZIP for LLM-assisted analysis and API workflows.
5 minutes to understand business workflows across a large, connected codebase.
Can you explain what the product does, how its files work together, and where a change could affect the workflow?
Explore smolagents reports ↗Used by developers across world-class firms such as Oracle, Dixon Technologies, DemocritusLabs, and more.
FROM DEVELOPERS WHO TRIED IT
Real repositories. Useful discoveries.
01 / UNDERSTANDING THE WORKFLOW
“The workflow and code maps helped me quickly trace relationships between routes, services, models, and database-related code.”
02 / FINDING A REAL BUG
“The Business Risk Findings report caught a real bug I didn't know about: my meditation-save endpoint always inserts a new row, but the model has a unique constraint on user + date, so a repeat save on the same day throws an unhandled error.”
START WITH YOUR REPOSITORY
One review.
Four practical outputs.
Install UnvibeCode and point it to your repository. No activation key or customer OpenAI API key is required.
py --version
py -m pip install --upgrade unvibecode
py -m unvibecode --help
py -m unvibecode review --repository "D:\path\to\repository"Requires Python 3.11+. Replace the example path with your repository path.
Supported languages
PythonJavaScriptTypeScriptRustPHPRubyHTML / CSS
For connected-code mapping and LLM-context preparation in v0.3.3. C, C++, Java, Go, C#, Kotlin, and Swift are detected but are not yet included in full connected-code analysis.
SAMPLE OUTPUTS / HUGGING FACE SMOLAGENTS
See what UnvibeCode finds in a real repository.
Explore the smolagents analysis: 30 workflows across 15 workflow areas, with supporting code and reusable context for your LLM.
Follow business workflows across connected files, inspect the supporting code, and take the relevant context into your AI assistant.
01 / BUSINESS WORKFLOWS
Follow the behaviour through the code
See how a multi-step agent uses a model, tools and memory to produce a final answer and an execution trace.
Explore business workflows →02 / CONNECTED CODE MAP
See which code belongs together
See how files and functions connect. Choose a file to explore its supporting code and prepare context for your AI assistant.
Explore connected code →03 / LLM REPOSITORY CONTEXT
Take the context into your LLM
Download the repository context to ask your AI assistant about workflows, code behaviour and changes.
Download LLM context →GO DEEPER
Guides for the work around the code.
Production RAG reliability
Explore parser routing, deterministic retrieval and calculation, knowledge graphs, source versioning, and evaluation.
Read the RAG guide ↗UnvibeCode RAG Review
Apply reusable review instructions, a checklist, and a runnable synthetic regression example to RAG work.
Explore the review skill ↗Pre-built business workflows
Explore the repository’s business workflow packs and contribution guidelines for adding a new pack.
Browse workflow packs ↗Related research: Evidence-Bound Factual Repair in Retrieval-Augmented LLM Answers.
UNVIBECODE VS OTHER CODEBASE TOOLS
Compare how you get to understanding.
UnvibeCode makes business and product behaviour the starting point: understand the workflow, locate its implementation, then assess the consequences of changing it.
| Tool | Strong at | Where UnvibeCode goes further |
|---|---|---|
| Probe | AST-aware code search, extraction, and code context for AI agents | Search and retrieval help locate code; UnvibeCode reconstructs the end-to-end business workflow that crosses those files and functions |
| Graphify | Building and querying a knowledge graph of code, documents, and relationships | A graph explains how things connect; UnvibeCode additionally reconstructs business workflows, edge cases, and evidence-backed business risks |
| PR-Agent / Qodo Merge | Reviewing pull requests, describing changes, and suggesting improvements around a diff | PR review starts from changed code; UnvibeCode reverse-engineers the existing repository and its business workflows beyond a single change set |
| Repomix | Packaging a repository into AI-friendly context for LLMs | Repository context gives an LLM source material; UnvibeCode additionally reconstructs workflow logic, state changes, edge cases, and business consequences |
| UnvibeCode | Connected code + business workflows + business logic + edge cases + evidence-backed business risks | The codebase is reviewed through the business workflows it implements, not only files, graphs, diffs, or context packages |
ENGINEERS ASK
Understand the workflow.
Explain the code.
What does UnvibeCode help engineers understand?
Explore a complex codebase’s business workflows in five minutes. See what the product does, which files work together, and how each workflow reaches its outcome—so you can explain the code you are changing without losing the bigger picture.
How do I understand AI-generated code with UnvibeCode?
Open the Business Workflow Map to see what your AI built. Pick a workflow, use the Connected Code Map to locate its implementation, then read Business Risk Findings before editing. You get a concrete route from product behaviour to the code that controls it.
How do I use UnvibeCode reports with ChatGPT, Cursor or Claude?
In the Connected Code Map, click a file and download its Optimal context. Attach it and the relevant HTML report to your assistant, or add them in Cursor. Ask: “Explain this workflow, cite its files, and show what my proposed change could affect.”
How does UnvibeCode uncover business workflows across files?
UnvibeCode maps files, symbols, imports and calls, then reconstructs the entry points, decisions, state changes and outcomes they implement together. For example, a subscription workflow connects renewal, cancellation and access removal—so an engineer can follow the product behaviour across its implementation.
How do I use all three maps and reports before changing code?
Business Workflow Map: identify the outcome your change affects. Connected Code Map: inspect the files and symbols involved. Business Risk Findings: inspect reported failure paths and their evidence. Turn those paths into tests before changing the workflow.
How is UnvibeCode different from Sourcegraph or a code graph tool?
Sourcegraph offers code search and AI answers; Graphify builds a queryable knowledge graph. UnvibeCode organises its review around business workflows, their connected implementation and evidence-backed risks. Start with what the product does, then identify the code and behaviour your change affects. Compare the tools.
How can I use prebuilt workflows with my coding assistant?
Copy a relevant pack—such as approvals, subscriptions or RAG—into your repository. Give its UNDERSTAND_CODE_SKILL.md and your code context to your assistant to trace the implementation; use TEST_WORKFLOW_SKILL.md to plan tests. Choose a workflow pack.
How can I use the AI engineering trade-offs guide?
Open the guide’s 40 production trade-offs, pick a decision relevant to your model, RAG or agent, and give it to your assistant with your requirements. Ask it to compare the options and justify a choice for your codebase. Explore AI engineering trade-offs.
Get UnvibeCode set up with your AI assistant.
Copy the setup question, open your assistant, and paste it to get started.
View setup question
How do I install UnvibeCode and get HTML reports and connected-code context ready for my complex codebase? Use https://github.com/FinanceFlash/unvibecode and its quick start. Ask for my operating system and repository path, give me the setup and review commands, and help me open 00_unvibecode_results.html. Show me how to download Optimal context from the Connected Code Map and prepare the relevant report files for ChatGPT, Cursor or Claude. Will UnvibeCode help me understand this repository? Assess its supported scope and limits, then use my actual reports to explain the main business workflows, their connected files and reported risks. Help me work toward a five-minute mental model; cite code evidence and identify anything the reports do not establish.
Unvibe complex code.
Trace what the business actually does.