<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje" -->

---
title: Giving Coding Agents a Map: Symbol-Level RAG With...
description: A prototype called nav aims to keep AI coding agents aligned with a codebase's existing conventions by building symbol-level context using tree-sitter for...
canonical: https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Giving Coding Agents a Map: Symbol-Level RAG With tree-sitter, Qdrant and SQLite | daily.dev
og:description: A prototype called nav aims to keep AI coding agents aligned with a codebase's existing conventions by building symbol-level context using tree-sitter for...
og:url: https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje
og:image: https://api.daily.dev/og/posts/sxUGbdeje.png
og:image:alt: Giving Coding Agents a Map: Symbol-Level RAG With tree-sitter, Qdrant and SQLite
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Giving Coding Agents a Map: Symbol-Level RAG With tree-sitter, Qdrant and SQLite

**[Grigorii Novikov](https://daily.dev/sources/qq7myouwwtfie0fvuckhu)** · [@oper18](https://daily.dev/oper18) · 1 min read · 1 upvotes · 0 comments

## Summary

A prototype called nav aims to keep AI coding agents aligned with a codebase's existing conventions by building symbol-level context using tree-sitter for parsing, Qdrant for vector storage, and per-branch SQLite for a knowledge graph. The post frames the problem as agents lacking context reproducing patterns from elsewhere (duplicate HTTP clients, inconsistent validation, bypassed repository interfaces) rather than just writing buggy code. It highlights three design choices: generating symbol summaries after building a project README rather than before, keeping the knowledge graph in per-branch SQLite instead of the vector store, and re-syncing the index on every prompt rather than on a schedule. Full technical details, including payload schema and hook internals, are linked externally. The author is soliciting failure modes from readers for a follow-up post focused on quantitative results.

## Content

An agent with no context doesn’t write bad code — it writes someone else’s code.
A second HTTP client. Validation inlined in the handler when the rest of the project does it in a service. A raw query in a codebase that’s been behind a repository interface for two years.
I built nav to stop that. The part that actually fixed it wasn’t vector search — it was three design decisions that sound wrong until you hit the problems they solve.
This is the first post in a series about the prototype.
It focuses on three design decisions behind nav:

Why every symbol summary is generated after a project README instead of before.
Why the knowledge graph lives in per-branch SQLite instead of the vector store.
Why the index syncs on every prompt instead of on a schedule.
[Full breakdown](https://dev.to/oper18/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-3ae5), including the payload schema and hook internals.
If you’re running agents against a mature codebase, how are you keeping them aligned with your project’s conventions? I’m collecting failure modes for Part 2, where I’ll dig into the numbers.

## Similar posts on daily.dev

- [I Replaced Vector DBs with Google’s Memory Agent Pattern for my notes in Obsidian](https://daily.dev/posts/i-replaced-vector-dbs-with-google-s-memory-agent-pattern-for-my-notes-in-obsidian-nfi4qzoyp) · Towards Data Science · 0 upvotes · 0 comments
- [The company that made RAG mainstream is now betting against it.](https://daily.dev/posts/the-company-that-made-rag-mainstream-is-now-betting-against-it--fykvhaqsi) · The New Stack · 2 upvotes · 0 comments
- [Google Newsletter Issue Agent-CLI Eval](https://daily.dev/posts/google-newsletter-issue-agent-cli-eval-zbtoft45i) · Daily Dose of Data Science \| Avi Chawla \| Substack · 3 upvotes · 1 comments
- [Building a RAG Pipeline for Semantic Code Search: A Developer Diary and Field Notes](https://daily.dev/posts/building-a-rag-pipeline-for-semantic-code-search-a-developer-diary-and-field-notes-wba9y41i0) · JetBrains · 4 upvotes · 0 comments
- [Local RAG for Agents: Integrating Private Knowledge Bases with Awesome-LLM-Apps](https://daily.dev/posts/local-rag-for-agents-integrating-private-knowledge-bases-with-awesome-llm-apps-gdqud2z89) · SitePoint · 0 upvotes · 0 comments

---

Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#qdrant](https://daily.dev/tags/qdrant), [#rag](https://daily.dev/tags/rag), [#sqlite](https://daily.dev/tags/sqlite)

[View this post on daily.dev](https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje","headline":"Giving Coding Agents a Map: Symbol-Level RAG With tree-sitter, Qdrant and SQLite","text":"A prototype called nav aims to keep AI coding agents aligned with a codebase's existing conventions by building symbol-level context using tree-sitter for parsing, Qdrant for vector storage, and per-branch SQLite for a knowledge graph. The post frames the problem as agents lacking context reproducing patterns from elsewhere (duplicate HTTP clients, inconsistent validation, bypassed repository interfaces) rather than just writing buggy code. It highlights three design choices: generating symbol summaries after building a project README rather than before, keeping the knowledge graph in per-branch SQLite instead of the vector store, and re-syncing the index on every prompt rather than on a schedule. Full technical details, including payload schema and hook internals, are linked externally. The author is soliciting failure modes from readers for a follow-up post focused on quantitative results.","url":"https://daily.dev/posts/giving-coding-agents-a-map-symbol-level-rag-with-tree-sitter-qdrant-and-sqlite-sxugbdeje","datePublished":"2026-09-15T19:51:47.845Z","dateModified":"2026-10-15T02:22:40.322Z","author":{"@type":"Person","name":"Grigorii Novikov","url":"https://daily.dev/oper18","image":"https://lh3.googleusercontent.com/a/ACg8ocKsP2ib2gClrCzbg4cVSNMJ6QiN2SU-k1QdjH--TzVB79WUi3eJ=s96-c","interactionStatistic":{"@type":"InteractionCounter","interactionType":{"@type":"EndorseAction"},"userInteractionCount":20}},"image":"https://media.daily.dev/image/upload/s--BCzQzZBR--/f_auto/v1789501908/posts/sxUGbdeje?_a=BAMAMicg0","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/sources/qq7myouwwtfie0fvuckhu","name":"Grigorii Novikov"}}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Grigorii Novikov","item":"https://daily.dev/sources/qq7myouwwtfie0fvuckhu"},{"@type":"ListItem","position":3,"name":"Giving Coding Agents a Map: Symbol-Level RAG With tree-sitter, Qdrant and SQLite"}]}
```

