update : Readme

README.md

@@ -1,84 +1,180 @@
 # apricot
-apricot is a self-hostable code agent , unlike tools that only look at diffs, apricot understands your entire codebase — reviews PR, write testcases 
---
-testing this works!!!
---
-# repo structure
+
+[![Typing SVG](https://readme-typing-svg.demolab.com?font=Fira+Code&duration=2800&pause=800&color=7C3AED&width=900&lines=Apricot+reviews+PRs+with+full-codebase+context;Apricot+generates+pytest+tests+from+diffs;Apricot+can+comment+on+PRs+and+push+generated+tests)](https://github.com/Neuropole/apricot)
+
+Apricot is a self-hosted AI code agent for pull requests.  
+It analyzes the latest Git diff, retrieves relevant code context from the repository, generates:
+
+1. **A structured review comment**
+2. **Pytest test cases** (saved to `tests/test_generated.py`)
+
+Then it can post the review on the PR and commit generated tests back to the PR branch.
+
+---
+
+## What it does
+
+- Reads recent code changes (`git diff HEAD~1 HEAD`)
+- Builds context from Python files in the repo
+- Chunks + embeds code with Sentence Transformers
+- Stores/query embeddings with ChromaDB
+- Uses Groq LLM for:
+  - review generation
+  - pytest generation
+- Publishes output to GitHub PR comments
+- Commits generated tests to the source PR branch
+
+---
+
+## Architecture
+
+```mermaid
+flowchart LR
+    A[PR Event / Local Run] --> B[agent.main]
+    B --> C[Diff Reader]
+    B --> D[Codebase Parser]
+    D --> E[Chunker]
+    E --> F[Embedder]
+    F --> G[Vector Store]
+    C --> H[Diff Embedding]
+    H --> G
+    G --> I[Relevant Context]
+    I --> J[Groq Review Generator]
+    I --> K[Groq Test Generator]
+    K --> L[tests/test_generated.py]
+    J --> M[GitHub PR Commenter]
+    L --> N[GitHub Committer]
+```
+
+### Runtime flow
+
+```mermaid
+sequenceDiagram
+    participant GH as GitHub Actions
+    participant Agent as agent.main
+    participant VS as Chroma Vector Store
+    participant LLM as Groq API
+    participant PR as Pull Request
+
+    GH->>Agent: Run on pull_request (opened/synchronize)
+    Agent->>Agent: git diff HEAD~1 HEAD
+    Agent->>Agent: Parse + chunk Python files
+    Agent->>LLM: Create embeddings (via local model)
+    Agent->>VS: Store code embeddings
+    Agent->>VS: Query relevant chunks using diff embedding
+    VS-->>Agent: Context chunks
+    Agent->>LLM: Generate review + pytest tests
+    Agent->>PR: Post review comment
+    Agent->>PR: Commit tests/test_generated.py
 ```
-ai-code-agent/
-│
-├── .github/
-│   └── workflows/
-│       └── agent.yml              # GitHub Actions pipeline
-│
-├── agent/                         # Core agent logic
-│   ├── __init__.py
-│   ├── main.py                    # Entry point (called by CI)
-│   │
-│   ├── config.py                  # API keys, configs
-│   │
+
+---
+
+## Project structure
+
+```text
+apricot/
+├── .github/workflows/agent.yml
+├── agent/
+│   ├── main.py
+│   ├── config.py
 │   ├── llm/
-│   │   ├── groq_client.py         # Groq API wrapper
-│   │   └── prompts.py             # All prompts (centralized 🔥)
-│   │
+│   │   ├── groq_client.py
+│   │   ├── prompts.py
+│   │   └── test_generator.py
 │   ├── indexing/
-│   │   ├── parser.py              # tree-sitter parsing
-│   │   ├── chunker.py             # split code into chunks
-│   │   ├── embedder.py            # embeddings logic
-│   │   └── vector_store.py        # ChromaDB wrapper
-│   │
-│   ├── retrieval/
-│   │   └── retriever.py           # top-k context fetch
-│   │
-│   ├── intent/
-│   │   └── infer.py               # 🔥 intent + property inference
-│   │
-│   ├── testgen/
-│   │   ├── generator.py           # test generation
-│   │   └── style_matcher.py       # detect pytest/jest etc.
-│   │
-│   ├── execution/
-│   │   ├── runner.py              # run tests (pytest etc.)
-│   │   └── coverage.py            # coverage analysis
-│   │
-│   ├── feedback/
-│   │   └── loop.py                # refine tests (self-improve 🔥)
-│   │
+│   │   ├── parser.py
+│   │   ├── chunker.py
+│   │   ├── embedder.py
+│   │   └── vector_store.py
 │   └── github/
-│       ├── commenter.py           # PR comments
-│       └── committer.py           # push test files
-│
-├── scripts/
-│   ├── setup.sh
-│   └── run_local.py               # run agent locally (debugging)
-│
-├── .agent/                        # persistent memory (IMPORTANT)
-│   ├── context.md                 # codebase summary
-│   ├── embeddings/                # cached embeddings
-│   └── intent_cache.json          # inferred properties
-│
-├── tests/                         # agent’s own tests
-│
-├── requirements.txt
-├── README.md
-└── .gitignore
+│       ├── commenter.py
+│       └── committer.py
+├── tests/
+│   ├── test_generated.py
+│   └── test_test_generator_formatting.py
+└── requirements.txt
+```
+
+---
+
+## Quick start (local)
 
+### 1) Install dependencies
+
+```bash
+pip install -r requirements.txt
+```
+
+### 2) Configure environment variables
+
+Create a `.env` file in the repository root:
+
+```env
+GROQ_API_KEY=your_groq_api_key
+GITHUB_TOKEN=your_github_token
+GITHUB_REPOSITORY=owner/repo
+PR_NUMBER=123
+GITHUB_HEAD_REF=feature-branch-name
 ```
-# end to end flow
+
+### 3) Run the agent
+
+```bash
+python -m agent.main
 ```
-PR opened →
-GitHub Actions triggered →
-    ↓
-1. Clone repo
-2. Load cached embeddings (.agent/)
-3. Parse changed files (diff)
-4. Embed only changed/new files
-5. Retrieve relevant context (RAG)
-6. Infer intent + properties 🔥
-7. Detect test framework
-8. Generate tests
-9. Run tests (pytest)
-10. Collect failures + coverage
-11. Feedback loop (optional refine) 🔥
-12. Commit tests OR comment on PR
-```
+
+---
+
+## GitHub Actions integration
+
+This repo includes `.github/workflows/agent.yml` that triggers on:
+
+- `pull_request` events: `opened`, `synchronize`
+
+The workflow:
+
+1. checks out code with full history
+2. sets PR number
+3. installs Python + requirements
+4. runs `python -m agent.main`
+
+Required secrets:
+
+- `GROQ_API_KEY`
+- `GITHUB_TOKEN` (GitHub-provided token works in Actions)
+
+---
+
+## Outputs
+
+Apricot produces:
+
+- **PR review comment** (bugs, improvements, suggestions)
+- **Generated tests file** at `tests/test_generated.py`
+
+---
+
+
+
+### CI pulse
+
+![Workflow Status](https://img.shields.io/github/actions/workflow/status/Neuropole/apricot/agent.yml?branch=main&style=for-the-badge)
+
+### AI activity banner (animated)
+
+![AI Activity](https://readme-typing-svg.demolab.com?font=JetBrains+Mono&duration=2200&pause=700&color=22C55E&width=900&lines=Diff+%E2%86%92+Context+Retrieval+%E2%86%92+LLM+Review;Diff+%E2%86%92+Context+Retrieval+%E2%86%92+Test+Generation;PR+Comment+%2B+Auto-Commit+of+Generated+Tests)
+
+---
+
+## Notes and current scope
+
+- Current implementation focuses on **Python codebases**
+- Test generation currently targets **pytest**
+- Diff source is the latest commit range (`HEAD~1..HEAD`)
+
+---
+
+## License
+
+MIT License (see `LICENSE`).

tests/test_generated.py

@@ -1,2 +1,27 @@
-def test_empty_diff():
+def test_get_pr_diff():
+    try:
+        subprocess.check_output(["git", "diff", "HEAD~1", "HEAD"], text=True)
+    except subprocess.CalledProcessError as e:
+        assert e.returncode != 0
+
+def test_generate_review():
+    review = generate_review("test_diff")
+    assert review is not None
+
+def test_post_comment():
+    post_comment("test_comment")
+    assert True
+
+def test_generate_pytest_test():
+    test_case = generate_pytest_test("test_diff")
+    assert test_case is not None
+
+def test_main_workflow():
+    # Simulate GitHub Actions workflow
+    os.environ["GROQ_API_KEY"] = "test_api_key"
+    os.environ["GITHUB_TOKEN"] = "test_token"
+    os.environ["GITHUB_REPOSITORY"] = "test_repo"
+    os.environ["PR_NUMBER"] = "123"
+    os.environ["GITHUB_HEAD_REF"] = "feature-branch-name"
+    main()
     assert True