frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Show HN: Vexp – Local-first context engine for AI coding agents

1•nicola_alessi•2h ago
I built vexp to solve two problems I kept hitting with AI coding agents (Claude Code, Cursor, etc.):

1. Token waste: agents read entire files linearly to understand a codebase. On a medium TypeScript project, a single query was consuming ~18k tokens of context when only ~2.4k were relevant.

2. Session amnesia: every new session starts from zero. The agent re-reads the same files, re-discovers the same architecture, re-traces the same dependencies.

How it works:

- tree-sitter parses the codebase and extracts symbols + relationships into a SQLite dependency graph - When an agent requests context, vexp serves a token-budgeted "capsule" containing only the relevant nodes: the target function, its callers, imported types, cross-file dependencies — not entire files - Session memory: every agent interaction is auto-captured as an observation linked to specific graph nodes. Next session, relevant observations surface automatically via MCP - Staleness detection: when code changes, observations linked to modified symbols are flagged stale. The agent sees "I learned X about this function, but it's changed since then" - Cross-repo: detects API contracts, shared types, and env contracts across multiple repos, all locally via workspace config

Architecture: native Rust binary + SQLite. No cloud, no embedding model, no external dependencies. Talks to agents via MCP (Model Context Protocol) stdio. Currently distributed as a VS Code extension but the MCP server is standalone.

The dependency graph is fully deterministic — AST structural analysis, not vector embeddings. Code relationships are facts (imports, calls, type references), not similarity scores. This means zero hallucination risk in the retrieval layer and trivial staleness detection (you know exactly which nodes changed).

Hybrid search for memory: FTS5 BM25 + TF-IDF cosine similarity + recency decay (7-day half-life) + graph proximity scoring - staleness penalty.

Supports 11 languages, 12+ MCP-compatible agents. Free tier includes context capsules + all session memory tools.

https://vexp.dev

Happy to discuss the architectur; the most interesting design decisions were around the scoring algorithm for pivot node selection and how to link observations to graph nodes without making the staleness detection too aggressive or too lenient.

Agent Harness for Coding with ClickHouse

https://clickhouse.com/blog/ai-powered-migraiton-from-postgres-to-clickhouse-with-fiveonefour
1•craneca0•25s ago•0 comments

I got tired of bouncing between AI tools, so I built 1 workspace for all of them

https://www.thetank.io/
1•schmommy•27s ago•1 comments

Show HN: As Level Chemistry Lab Simulator

https://github.com/nsriram/chem_lab
1•sriram_iyengar•43s ago•0 comments

Mato – a Multi-Agent Terminal Office workspace (tmux-like)

https://mato.sh
1•chepy•1m ago•0 comments

Show HN: Golos – Ultra Lightweight Free Wispr Flow Alternative

https://github.com/basilysf1709/golos
1•basilyusuf1709•2m ago•0 comments

Show HN: BVisor – An Embedded Bash Sandbox, 2ms Boot, Written in Zig

https://github.com/butter-dot-dev/bVisor
1•edunteman•3m ago•0 comments

Transitive Core Concepts: Full-Stack Robotic Packages

https://transitiverobotics.com/blog/design-full-stack-packages/
1•chfritz•3m ago•1 comments

Show HN: OmniGlass – An open-source, sandboxed Visual Action Engine

https://github.com/goshtasb/OmniGlass
1•goshtasb•7m ago•0 comments

25 Years of Eggs

https://www.john-rush.com/posts/eggs-25-years-20260219.html
1•derek•7m ago•0 comments

Multiple formation pathways for amino acids based on asteroid sample isotopes

https://www.pnas.org/doi/10.1073/pnas.2517723123
1•PaulHoule•7m ago•0 comments

Show HN: Personal Vault – Own your personal context, let AI agents query it

https://github.com/lovincyrus/personal-vault
1•lovincyrus•8m ago•0 comments

Show HN: I made a hub to fix how founders and investors connect online

https://startupa.ge/
1•guglielmomave•8m ago•0 comments

Show HN: I Built Captcha for Agents

https://agent-captcha.dhravya.dev/
1•dhravya•8m ago•0 comments

Tariff Refund Calculator

https://tariffs.flexport.com/refunds
1•thedogeye•9m ago•0 comments

User gains control of over 6,700 DJI robot vacuums with help from Claude Code

https://www.tomshardware.com/tech-industry/cyber-security/user-accidentally-gains-control-of-over...
1•gradus_ad•10m ago•0 comments

Rust Doesn't Have Named Arguments. So What?

https://thoughtbot.com/blog/rust-doesn-t-have-named-arguments-so-what
1•romac•10m ago•0 comments

Show HN: DealLedger – An open ledger of every business for sale in America

https://dealledger.org
1•jsosville•11m ago•0 comments

Native apps with Tauri, back end with Bun

https://codeforreal.com/blogs/using-bun-or-deno-as-a-web-server-in-tauri/
1•zbycz•11m ago•0 comments

Surprisingly Frugal Hobbies

https://cmart.blog/surprisingly-frugal-hobbies/
1•zdw•11m ago•0 comments

OpenAI lands multiyear deals with consulting giants in enterprise push

https://www.cnbc.com/2026/02/23/open-ai-consulting-accenture-boston-capgemini-mckinsey-frontier.html
2•Betelbuddy•12m ago•0 comments

Venture Fraud

https://www.nber.org/papers/w34868
1•bikenaga•12m ago•1 comments

Effective Agents

https://medium.com/@NAI_Journal/how-i-built-a-multi-agent-development-team-with-claude-code-mcp-a...
1•nola-a•12m ago•0 comments

I made a retro 90s TCG

http://arcana.online/
4•D_ashe•13m ago•1 comments

The Flawed V02 Max Craze

https://erictopol.substack.com/p/the-flawed-v02-max-craze
1•brandonb•13m ago•0 comments

Ask HN: Is there anyone working at Meta who could help with a stolen FB account?

2•snicky•14m ago•0 comments

Show HN: AgentDbg - local-first debugger for AI agents (timeline, loops, etc.)

https://github.com/AgentDbg/AgentDbg
2•z-a-f•15m ago•1 comments

HubSpot Acquires YouTube-Based Media Brand Starter Story

https://www.adweek.com/media/hubspot-media-acquires-starter-story-youtube/
1•lasgawe•15m ago•0 comments

ReadInSync: Personalized, grouped news summaries to reduce noise

https://readinsync.com/
1•agnim25•16m ago•0 comments

Sam Altman would like to remind you that humans use a lot of energy, too

https://techcrunch.com/2026/02/21/sam-altman-would-like-remind-you-that-humans-use-a-lot-of-energ...
1•trinsic2•17m ago•2 comments

Deplatform Yourself

https://pluralistic.net/2026/02/23/goodharts-lawbreaker/
8•hn_acker•17m ago•0 comments