readme

[hemanth-1321]

Summary by CodeRabbit

• New Features

  • Enhanced multi-language semantic code graphs with clearer functions, classes, imports, calls, decorators, and unique labels for anonymous items.

• Performance

  • Faster vector search setup and batch context storage for improved processing throughput.

• Documentation

  • Expanded README with high-level system design, workflow overview, and key characteristics.
  • Added an architecture diagram.

• Tests

  • Introduced a PR-focused code analysis script to parse files and build AST/semantic graphs.

• Chores

  • Updated ignore rules; removed a deprecated import-check utility.

• Style

  • Minor logging message cleanup and formatting adjustments.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Preview	Comments	Updated (UTC)
code-daddy	Ready	Preview	Comment	Oct 15, 2025 9:45am

[coderabbitai]

Caution

Review failed

The pull request is closed.

Walkthrough

Updates ignore rules and documentation; deletes a client script; adds an Excalidraw diagram; tweaks minor logs/formatting. Removes “learnings” vector search usage, simplifies vector_tool, and introduces batch-oriented Qdrant storage with HNSW config. Overhauls semantic graph construction to be language-agnostic. Adds a Tree-sitter-based analysis/test script.

Changes

Cohort / File(s)	Summary
Repo metadata & docs \.gitignore, README.md, codedaddy.excalidraw	Adjust ignore patterns (remove lib/, lib64/, add parts/). Replace README header with a high-level system design section. Add new Excalidraw diagram JSON.
Removed client utility client/checl.ts	Delete TypeScript import-checking script.
Agent flow simplifications server/agentic/agents/aggregator_agent.py, server/agentic/agents/nodes.py, server/agentic/utils/vector_tool.py	Remove comments in aggregator. In nodes: drop search_learnings_tool usage; set static learnings string. In vector_tool: remove learnings search function; adjust search to use full content; minor no-op print.
Qdrant vector storage (batch + HNSW) server/agentic/utils/qdrant_db.py	Switch to HNSW-based collection config (COSINE distance). Replace per-PR upserts with batch prepare_and_store_context(pr_contexts: List[Dict]). Create points list and upsert in one call.
Minor server behavior tweaks server/agentic/main.py, server/routes/webhook.py	Change error log message format in process_ai_job; add a blank line in webhook route.
Graph construction overhaul server/worker/services/graph_utils.py	Add LANG_RULES; refactor AST/semantic graph builders to a language-agnostic, rules-driven approach. Add unique IDs for anonymous defs; introduce file anchor node; unify edges: contains/imports/uses_import/has_decorator/calls.
New analysis script test.py	Add Tree-sitter pipeline: language rules, parsing, AST/semantic graph building, and a PR-processing workflow that clones and analyzes changed files.

Sequence Diagram(s)

sequenceDiagram
  autonumber
  participant Caller as Caller
  participant QDB as QdrantDB Utils
  participant Qdrant as QdrantClient

  Caller->>QDB: prepare_and_store_context(pr_contexts: List[Dict])
  Note right of QDB: Build HNSW VectorParams (COSINE, m, ef_construct)
  QDB->>Qdrant: Create collection if missing (HNSW config)
  QDB->>QDB: Map contexts -> PointStruct[]
  QDB->>Qdrant: Upsert(batch Points)
  Qdrant-->>QDB: Upsert result
  QDB-->>Caller: Success/summary

Loading

sequenceDiagram
  autonumber
  participant SRC as Source Code
  participant Parser as Tree-sitter Parser
  participant Graph as Semantic Graph Builder

  SRC->>Parser: parse(lang)
  Parser-->>Graph: AST
  Graph->>Graph: Identify defs/imports/calls via LANG_RULES
  Graph->>Graph: Create nodes (file anchor, defs, imports, calls)
  Graph->>Graph: Connect edges (contains, imports, uses_import, has_decorator, calls)
  Graph-->>Caller: Semantic graph
  Note over Graph: Anonymous defs get UUID-suffixed labels

Loading

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~60 minutes

Possibly related PRs

• Test error #12 — Introduces a Tree-sitter-based PR parsing and code-graph construction pipeline with similar functions and language maps.

Poem

A rabbit taps keys with a gentle thrum,
Batch vectors hop—one upsert, not some.
Graphs weave trails where functions play,
Calls and imports map the way.
Learnings rest, for now they sleep—
While HNSW burrows fast and deep. 🐇✨

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch worker

---

📜 Recent review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 7c8ee21 and 9a80bab.

📒 Files selected for processing (12)

• .gitignore (1 hunks)
• README.md (1 hunks)
• client/checl.ts (0 hunks)
• codedaddy.excalidraw (1 hunks)
• server/agentic/agents/aggregator_agent.py (0 hunks)
• server/agentic/agents/nodes.py (2 hunks)
• server/agentic/main.py (1 hunks)
• server/agentic/utils/qdrant_db.py (2 hunks)
• server/agentic/utils/vector_tool.py (1 hunks)
• server/routes/webhook.py (1 hunks)
• server/worker/services/graph_utils.py (3 hunks)
• test.py (1 hunks)

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}