Vexp - Exclusive Lifetime Deal: $49 once vs $228/yr | AppSumo
Share Vexp
Vexp
Build a local dependency graph and serve precise context to AI coding agents across VS Code and CLI
10% off ends in
17
Hours
:
29
Mins
:
44
Secs
-51%$49
$100
Select a planView plan details
License Tier 1:$49$100
Buy now
- Lifetime access
- Refundable up to 60 days
- Plus members are covered by our We Got Your Back guarantee
Chosen by AppSumo for their potential and innovation
Your AI agent just re-read half your codebase to add one function. Good thing Vexp can build a map of your codebase and serve your agent exactly what it actually needs to get it right the first time.
TL;DR
- Serve precise, dependency-aware context to AI coding agents across 11 tools in VS Code and CLI
- Run fully offline with your code never leaving your machine—no network calls, no cloud storage
At-a-glance
Integrations
GitHub, Microsoft VSCode
Best for
Developers, SaaS, Solopreneurs
10% off ends in
17
:
29
:
44
Vexp
Stop paying for tokens your agent wastes
10% off ends in
17
Hours
:
29
Mins
:
44
Secs
Give AI coding agents precise, dependency-aware context from a local graph
Build a local dependency graph
- Parse your codebase into a graph of functions, classes, and types using tree-sitter indexing across 30+ languages
- Return only the files and symbols an agent actually needs —bound to your token budget, never an empty capsule
- Trace call chains and import paths across files to understand exactly what each piece of code depends on
- Enable vexp-devmind to preprocess prompts locally, an optional on-device AI model that compresses context before it reaches the cloud agent, with zero data leaving your machine
Keep your code off the cloud
- Run entirely on your machine with zero network calls —your source code never leaves your local environment
- Store all indexes in .vexp/index.db, a local SQLite file that stays in your repo
- Filter secrets by default so sensitive credentials are never included in any context capsule
Connect any agent, any workflow
- Integrate with Claude Code, Codex, Cursor, Windsurf, and 8 more AI coding agents via a single MCP server
- Pick up exactly where you left off. Vexp remembers what your agent explored in previous sessions and flags stale context automatically when code changes
- Expose 11 specialized MCP tools for context search, impact analysis, logic tracing, and more
- Install via the VS Code extension or standalone CLI, and Vexp generates your agent's MCP config automatically
Set up in minutes, index in seconds
- Run the Vexp setup to configure your indexes, agents, and git hooks in a single guided command
- Add and connect multiple repositories to one workspace for cross-repo context queries
- Access setup, daemon controls, and local LLM configuration from the same VS Code actions panel or CLI menu
Choose the plan that’s right for you
Feel secure in your purchase with AppSumo's60 day money-back guarantee.
| ##### License Tier 1 $49$100 Buy now Lifetime access |
##### License Tier 2 $119$240 Buy now Lifetime access |
Recommended ##### License Tier 3 $199$1,800 Buy now Lifetime access |
##### License Tier 4 $399$3,600 Buy now Lifetime access |
|
|---|---|---|---|---|
| Nodes | 10,000 | 50,000 | 100,000 | Unlimited |
| Seats | 1 | 1 | 5 | 10 |
| Repo per workspace | 1 | 3 | Unlimited | Unlimited |
| Shared workspace & git index | ||||
| Priority paths & analytics | ||||
| Intent detection & codelens | ||||
| LSP bridge | ||||
| Session memory | ||||
| Hybrid search | ||||
| Feedback loop | ||||
| Passive observation | ||||
| All 12 MCP tools | ||||
| Git manifest sync |
Refundable up to 60 days
Plus members are covered by our
We Got Your Back guarantee
Deal terms & conditions
Open Deal terms & conditions
Get lifetime access to Vexp.
All future Pro (Tiers 1, 2), Team (Tiers 3, 4) plan updates.
If the plan name linked to your deal changes, it will automaticallytransfer to the new plan with all updates.
Activate your license within 60 days—you can upgrade while the deal is live but can only downgrade within 60 days of purchase.
\ See Vexp](https://vexp.dev/)
Founded February 26, 2026
🇮🇹Milan, Italy
1-10
Growth
Bootstrapped
We use PandaDoc to confirm each founder's identity.
Targeted AI code assistance
Vexp emerged to address the challenge of overwhelming code context in AI-driven development.
Its founders observed that AI coding agents often struggled with large and complex codebases, sending bloated context to language models. To solve this, they built a local-first dependency graph engine that surfaces only the code relevant to a given task.
By operating entirely on the developer’s machine, Vexp reduces token usage by up to 70%, accelerates code generation, and safeguards code privacy.
Founder
Questions & reviews
Search
Reviews29Questions107FAQs15From the founders1
Have a question?
Here you'll find a curated collection of frequent questions and answers, designed to provide you with quick, clear information about the product.
For all other questions refer to the questions tab
Helpful links
Open Helpful links
What is Vexp and how does it enhance AI coding agents?
Open What is Vexp and how does it enhance AI coding agents?
Vexp is a local-first context engine designed for AI coding agents. It builds a live dependency graph of your codebase, enabling the agent to access only the relevant code for a given task. This approach reduces token usage by up to 70%, leading to more efficient and accurate code generation.
How do I install Vexp for use with VS Code?
Open How do I install Vexp for use with VS Code?
To install Vexp in VS Code, open the Extensions panel (Ctrl+Shift+X or ⌘⇧X), search for 'vexp', and click 'Install'. The extension includes the native engine for your platform, requiring no additional setup.
Which code editors and tools does Vexp support?
Open Which code editors and tools does Vexp support?
Vexp auto-configures itself for the major AI coding agents, including VS Code, Cursor, Claude Code, Codex, Windsurf, Zed, and others. Installation is one click in the extension, or via the CLI, and Vexp writes the MCP configuration for your agent automatically. Remote-SSH workflows (including WSL2) are supported. For agents not on the auto-configured list, Vexp can still connect via its MCP server with a manual configuration.
What is a "node" and how do I know how many I need?
Open What is a "node" and how do I know how many I need?
A node is a single piece of code that Vexp maps: a function, a class, or a type. A project (a repo) is made up of many nodes, so the node count measures how large a project is, not how many projects you have. Separate, unrelated projects are kept as separate workspaces, and individual projects are unlimited on every tier. As a rough guide: most solo developers working on a few small-to-mid projects fit comfortably in Tier 1 or 2, while a handful of large repos, or many repos linked together into one workspace, call for Tier 3. The easiest way to size it is to run the free tier on your project and watch the node count.
What happens to my node budget if I remove a project?
Open What happens to my node budget if I remove a project?
Node usage reflects what's actively connected, so it's never permanently consumed. If you remove a repo from a workspace (by updating the workspace config and reindexing), those nodes are freed and can be reassigned to new projects. A standalone project you stop using doesn't count against your budget.
What operating systems are supported by Vexp?
Open What operating systems are supported by Vexp?
Vexp supports multiple operating systems, including Linux x64 (Ubuntu 20.04+, Debian 11+, RHEL/Rocky 8+, WSL2), Linux arm64, macOS Intel and Apple Silicon (macOS 11 or later), and Windows x64 (Windows 10 or later).
How does Vexp handle multi-repo workspaces?
Open How does Vexp handle multi-repo workspaces?
Vexp can index multiple repositories as a unified workspace, detecting relationships between them. This is useful for monorepos split across repos, frontend and backend pairs, and microservices. Setup involves creating a '.vexp/workspace.json' file in your primary repo to define the workspace configuration.
Which programming languages does Vexp support?
Open Which programming languages does Vexp support?
Vexp supports 27 programming languages, so most common stacks are covered. That includes TypeScript, JavaScript, Python, Go, Rust, Java, C#, C, C++, Ruby, PHP, Kotlin, Swift, Scala, Dart, Bash, Elixir, R, Lua, and more, plus Terraform/HCL and Dockerfile, and it indexes Markdown docs too. Standard web stacks (JS/TS, HTML, including Salesforce Lightning Web Components) work out of the box. The main gap worth calling out is Salesforce Apex, which isn't supported yet. If there's a specific language you need, let us know, demand shapes what we add next.
Does Vexp offer a local LLM for enhanced processing?
Open Does Vexp offer a local LLM for enhanced processing?
Yes, Vexp provides an optional on-device AI layer that adds smart preprocessing, intent detection, query expansion, session summarization, re-ranking, and anti-pattern nudging. This local LLM runs entirely on your machine, offering additional token savings of approximately 30%.
How does Vexp reduce token usage in AI coding agents?
Open How does Vexp reduce token usage in AI coding agents?
Vexp reduces token usage by building a live dependency graph of your codebase and serving the AI agent only the code relevant to the task. This targeted approach minimizes the amount of code processed, leading to significant token savings.
How does licensing work across multiple machines?
Open How does licensing work across multiple machines?
Licenses are per device: one seat covers one machine. If you use both a Mac and a Windows machine and want Vexp active on both, that's two devices, so you'd need two seats (Tier 3). Plan your tier accordingly if you regularly work across multiple machines.
What are the benefits of using Vexp with AI coding agents?
Open What are the benefits of using Vexp with AI coding agents?
Using Vexp with AI coding agents offers several benefits, including reduced token usage (up to 70% savings), improved code generation accuracy, faster processing times, and enhanced privacy by operating entirely on your local machine without requiring an internet connection.
How does Vexp integrate with Claude Code?
Open How does Vexp integrate with Claude Code?
Vexp integrates with Claude Code by acting as an MCP server. After installing Vexp and initializing your project, you configure Claude Code to use Vexp as an MCP server by adding specific settings to the '.claude/settings.json' file. This setup enables Claude Code to utilize Vexp's context-aware capabilities.
Is an internet connection required to use Vexp?
Open Is an internet connection required to use Vexp?
No, Vexp operates entirely on your local machine. It does not require an internet connection, ensuring that your code remains private and secure.
Expected to respond within 24 hours
Milan, Italy
Log in to ask a question
Sort by:
Latest reply
Showing 5 of 107 questions
gustavo615](/content/profile/gustavo615/index.html)
gustavo615
Sep 21, 2026
Q: Question about long-term access to the Lifetime Deal
Hi! I have a few questions about the lifetime license, just to better understand how it works long term:
1. If the company were ever to permanently cease operations, would an already activated Tier 2 installation continue to work indefinitely?
2. Does the lifetime license key have any internal expiration date?
3. If the licensing servers were ever shut down, would there be a final unlocked or fully offline version available for lifetime users?
4. In that scenario, is there any plan to open-source the core product or release the licensing mechanism so existing lifetime users could continue using it?
Thanks — I’m just trying to understand the long-term continuity of the lifetime plan before purchasing.
Founder Team
Nicola_Vexp
Sep 22, 2026
A: Fair questions before buying a lifetime deal, and I'd rather answer them straight than reassure you vaguely.
1 and 3: if vexp ever ceased operations, I would release a final version that works with no server-side key verification for every existing lifetime customer. That's a commitment, and one I'd make sure is honored.
2: yes, today license keys have a monthly expiry and renew automatically through our licensing server. That's precisely why the commitment above matters: in a shutdown scenario, I'd make sure that dependency is removed so your installation keeps working indefinitely.
4: there's no plan to open-source the core. The commitment above is what addresses the real concern behind the question, which is that lifetime customers keep a working product regardless of what happens to the company.
Worth adding that the engine itself is local-first: indexing and querying all happen on your machine with zero network calls. The licensing server is the only moving part, and that's the one I'd take care of.
Comment (4)
Share Vexp
Helpful?
2
Log in to join the conversation
Ejay PLUS
And can you speak to what you could/would do if you were bought out by one of these big companies, please? Cause realistically, this is what would happen since it's running local, you don't have much overhead, correct?
Founder
Verified purchaser
Honest answer, including the part I can't promise. No acquisition conversations are happening, and if one ever did, lifetime licenses transfer with the assets, an acquirer buys the commitments along with the product. But no founder can fully bind a future owner, and anyone claiming otherwise is overselling.
Founder
Verified purchaser
The real protection isn't a promise though, it's the architecture. vexp runs entirely on your machine with zero network calls. An acquirer could stop developing it or take the site down, and your installed copy keeps working on your code. Worst realistic case is it stops improving, not that it stops running, which is a very different risk from a cloud tool.
Founder
Verified purchaser
And your overhead point actually cuts the other way: no cloud bill scaling with usage is exactly why a lifetime deal is sustainable here, and why there's no pressure to find an exit.
StackPilot](/content/profile/StackPilot/index.html)
StackPilotPLUS
Sep 21, 2026
Q: Can Vexp help non-developers maintain AI-generated code?
Hi Vexp Team,
I'm evaluating Vexp for maintaining applications built with AI-assisted development tools.
Could you please clarify:
Can Vexp analyze an existing codebase generated by an AI app builder, or must the project be developed directly inside VS Code?
Does Vexp work with common JavaScript/TypeScript frameworks and projects hosted outside the local development environment?
Can a non-developer use Vexp alongside an AI coding assistant to identify dependencies, understand the impact of proposed changes, and reduce the risk of breaking existing functionality?
Does Vexp provide its own AI coding assistance, or does it only supply code context to external AI agents?
Thank you!
Founder Team
Nicola_Vexp
Sep 21, 2026
A: Existing AI-generated code: yes. vexp indexes whatever code is on your machine, regardless of who or what wrote it, and it doesn't need to have been built in VS Code (there's also a CLI). The one requirement is that the files are local: if your app came from a cloud builder like Lovable or Bolt, export it via GitHub first.
Frameworks and hosting: JS and TS are fully supported, plus Vue, Svelte and Astro. vexp works on source code, not live deployments, but it runs over Remote SSH if your code sits on a remote machine.
Non-developers: yes, and it's a strong fit. Ask your agent "what breaks if I change this?" and it gets the real dependency chain instead of guessing. There's also a completion check that flags broken imports, unupdated dependents and affected tests before you ship.
Its own AI: no. vexp supplies context to your existing agent (Claude Code, Cursor, Codex, Copilot) rather than writing code itself.
Free tier lets you try it on one of your projects first.
Share Vexp
Helpful?
0
Log in to join the conversation
mastro100s](/content/profile/mastro100s/index.html)
mastro100sPLUS
Sep 19, 2026
Q: Daily_usage integrity check failed
- Your current error: Vexp reports an integrity-check failure with a recorded count of just 2, then treats the quota as exhausted. This does not demonstrate that you used 1,500 calls.
The immediate blocker is Vexp’s usage-validation error. The exact message to send their support is: ... pipeline_skeleton (count=2) on Tier 4, bundled core version 3.2.2.
Why?
Founder Team
Nicola_Vexp
Sep 20, 2026
A: You're right, and thanks for quoting the exact log line: two calls, not 1,500.
vexp signs every daily-usage row, so a counter that has been edited by hand counts as spent for the day. That is the anti-tamper you want behind a lifetime deal. In your case two calls fired in the same agent turn (run_pipeline and get_skeleton together, which is normal) raced each other on that signature: the row ended up holding two calls with the signature for one, and the daemon treated it as tampered. The "1500/1500" is how that state is displayed, not real usage.
Fixed: the counter and its signature are now written in one atomic transaction, so this cannot happen again. It ships in 3.2.4 today. Meanwhile the counter is keyed by UTC date, so a day that tripped heals itself at midnight UTC, and Tier 4's 1,500 calls per day is a ceiling meant for automated fleets. Interactive use, even a heavy day, does not get near it.
Share Vexp
Helpful?
0
Log in to join the conversation
Blueyetii](/content/profile/Blueyetii/index.html)
BlueyetiiPLUS
Edited Sep 20, 2026
Q: I am a vibe coder putting out 5-10 mobile apps a month what tier you recommend me getting tier 3 or tier 4
Founder Team
Nicola_Vexp
Sep 20, 2026
A: Tier 3, comfortably, and I'd be surprised if you ever felt the ceiling.
Here's why. The node limit applies per workspace, not as a running total, so shipping 5 to 10 apps a month isn't the constraint at all. Each app is its own workspace and starts fresh against the limit. What matters is the size of a single project, and mobile apps are small in these terms: for reference, a 472-file project measured around 4,400 nodes, and a 2,000-file legacy codebase came in at 55,000. A typical mobile app sits in the low thousands. Tier 3 gives you 100k per workspace, so you'd need a genuinely enormous single app to get close.
Tier 3 also gives you unlimited repos per workspace and 5 device seats, which covers most solo setups with room to spare.
The only real reasons to go Tier 4 are needing more than 5 devices, or single projects big enough to pass 100k nodes. Neither sounds like your situation. One thing worth knowing though: tier upgrades only exist while the AppSumo deal is live, so if you think you'll want 6+ machines later, that decision has to happen now. Otherwise Tier 3 is the right call.
Share Vexp
Helpful?
1
Log in to join the conversation
112989862057960397719](/content/profile/112989862057960397719/index.html)
112989862057960397719
Sep 19, 2026
Q: email assistenza?
I purchased Tier 2 and am quite satisfied with the product, which I use with Trae. However, today I see the following message in the Trae IDE: "vexp: Failed to start daemon: Error: spawn EPERM. Check .vexp/vexp.log for details." The last few lines of the log read: "--- vexp daemon start at 2026-09-19T08:03:39.925Z ---
[daemon] start() failed: Error: spawn EPERM
[daemon] daemon died during startup — respawning in 30000ms (attempt 5/5)
[daemon] respawn failed: Error: spawn EPERM
". How can I contact support? I can't find the email address anywhere, and I do NOT use Outlook.
Thx
Founder Team
Nicola_Vexp
Sep 19, 2026
A: Sorry about that, and the email is staff@vexp.dev, write to me there directly (it's also linked under Contact Support in the account portal, but it's not as findable as it should be, that's on us).
On the error itself, spawn EPERM is a permissions problem rather than a vexp bug: something on your machine is blocking the engine binary from being executed. The usual culprit is antivirus or endpoint protection quarantining it, which is common on Windows with corporate security tools. Worth checking your antivirus quarantine or blocked-app list for vexp-core and allowing it.
When you email me, run vexp doctor --report first and attach the .vexp/vexp-report.md it produces. It has everything I need to pinpoint it (doctor output, logs, config, secrets masked), and it saves us several rounds of back and forth. If the daemon won't start at all, plain vexp doctor still works without it.
Send it over and I'll get you running.
Comment (2)
Share Vexp
Helpful?
0
Log in to join the conversation
Verified purchaser
Nicola, thanks for the lightning-fast reply. I'll run all the checks first, and if I manage to fix it, I won't even need to bother you!
Nicola replied less than 2 minutes after my request! I've never seen anything like it! Top-notch service!
Founder
Verified purchaser
You're very welcome, and thank you for the kind words. Honestly, it's never a bother. Reach out whether you fix it or not, if you do solve it I'd love to know what it turned out to be, because that helps me spot patterns and stop the next person hitting the same thing.
And that's the standing offer: me and the team are here for you, always. No ticket queues, no runaround. Glad to have you with us
Vexp is praised for its efficient token savings, precise context provision to AI agents, and local-first privacy. However, some users have reported minor issues with the licensing management and initial setup. With an overall rating of 4.5 and a 60-day money-back guarantee, it's worth giving Vexp a try for those seeking a reliable AI coding tool that optimizes coding sessions and enhances workflow efficiency.
AI-powered summary of customer reviews
Log in to leave a review
Ratings
Ratings
4.5
29
Taco ratings
5 tacos
234 tacos
23 tacos
12 tacos
21 taco
1
Log in to leave a review
Sort by:
Newest
Taco rating:
All tacos
Showing 5 of 29 reviews
3466f85e331c4277ab15fc4b2cd8bfc8 PLUS
Verified purchaser
Deals bought: 230Member since: Feb 2017
Sep 20, 2026
Helpful Cursor setup for Windows and Remote SSH users
I purchased Vexp through AppSumo and have only just completed the initial setup, so this is not yet a long-term performance review. I still wanted to share my experience because the Cursor setup, especially for Windows and Remote SSH users, may be helpful to the community.
I installed the Vexp extension in Cursor, activated my Lifetime Tier 3 license, configured the project integration, and created the initial code index.
One question I had before installing was whether the extension download was official. Cursor did not initially show Vexp in the built-in Extensions search, and the Open VSX page displayed an “unverified namespace” notice. I contacted Vexp support directly.
Nicola confirmed that Vexp.vexp-vscode, published by nicolalessi on Open VSX, is the official Vexp extension. He explained that the unverified namespace status is an Open VSX administrative verification process that is still pending; it does not mean the package is unofficial or unsafe. He also confirmed that it is the same package distributed through the Microsoft Visual Studio Marketplace.
The official download sources are:
Open VSX: https://open-vsx.org/extension/Vexp/vexp-vscode
Microsoft Visual Studio Marketplace: https://marketplace.visualstudio.com/items?itemName=Vexp.vexp-vscode
For Cursor on Windows, the recommended approach is to open Extensions with Ctrl+Shift+X, search for vexp, and install:
Vexp — AI Context Engine for Claude Code, Cursor & Copilot
Publisher: Vexp / nicolalessi
If Cursor does not find it in the extension search, the Open VSX page provides the official fallback download and VSIX installation route.
Another important thing I learned: if you work with Cursor through Remote SSH, Vexp needs to be configured inside the remote SSH environment as well. Installing it only in local Windows Cursor is not enough. In the remote project, I ran Vexp setup for Cursor, started indexing, enabled the Vexp MCP server in Cursor, and then opened a new Agent chat.
After that, Cursor was able to use the project context and identify relevant files, functions, and dependencies much more precisely. I also increased the Vexp indexing limit for one large source file so it could be included in the code graph.
So far, the setup is working well both locally and on my SSH server. I have not used it extensively in daily development yet, but the initial installation and integration experience has been positive. I am looking forward to seeing how much time and context it saves in larger projects.
Helpful tip: After changing the MCP configuration or updating the extension, reload the Cursor window and open a new Agent chat.
Read less
Founder Team
Nicola_Vexp
Sep 20, 2026
Thank you, and not just for the rating: you've written the clearest explanation of the Open VSX situation that exists anywhere right now, including ours. That namespace notice trips up a lot of Cursor and Windsurf users, and having it laid out by someone who actually verified it themselves is far more convincing than us saying it.
The Remote SSH section is genuinely useful too. It catches people out, because the agent and the engine have to live on the same side of that boundary, so installing locally on Windows while the project lives on the remote leaves vexp with nothing to index. Your write-up covers it properly, including the reload and new Agent chat step, which is the detail most people miss.
Good call on raising the file size limit for that large source file as well, oversized files get skipped rather than partially indexed, so that one setting can matter more than it looks.
Looking forward to hearing how it goes once you've had it in real use on bigger projects. And thanks for taking the time to write this for other people rather than just for yourself.
Helpful?
2
Share Vexp
Verified purchaser
Deals bought: 34Member since: Apr 2025
Sep 19, 2026
Great Tool, But "Team Plan" Has a Severe Security & Management Flaw (Read Founder Reply)
PRO:
\* Excellent Functionality: Vexp is fantastic for saving tokens, making code cleaner, and improving workflow accuracy.
\* Local-First Privacy: I highly value that it runs locally and keeps my private code secure.
CON:
While the tool itself works beautifully, the way Vexp handles license keys for its Tier 4 (10 Seats) plan is a major operational and security risk.
The product page explicitly sells this tier as having "10 seats." In the software industry, "seats" implies individual access control for different users. However, Vexp doesn't actually give you 10 individual seats; it issues one single license key for everyone to share. The major issue is that removing a device from the dashboard frees up a slot, but it does not invalidate the key. Because there is no key rotation or revocation feature, anyone who has ever seen that key can re-activate it on a new device at any time. If an employee leaves or a key is leaked, unauthorized people can repeatedly hijack the paid slots you bought.
------------------------------
## Response to the Founder's Reply:
The founder’s defense is that this is a "licensing gap/revenue loss for him" rather than a security breach for the customer because local code files aren't exposed. This completely misses the point of asset security.
\* My Paid Slots Are the Asset: I am not worried about my files being accessed; I am worried about the integrity of the slots I paid for. If a leaked key allows outsiders to flood my account and lock out my actual users, my paid service is disrupted. That is a security and access issue for my business.
\* The "Local-First" Excuse Doesn't Hold Up: Being local-first doesn't mean the licensing layer has to be defenseless. The app clearly communicates with vexp.dev to verify slots. If the server allowed us to rotate the key, the local app would simply reject the old key the next time it validates.
\* Offboarding vs. Hacking: The founder argues that if a machine is compromised, a license key is the least of my worries. This ignores the most common business scenario: standard employee offboarding. When a worker leaves, we revoke their access and credentials. But with Vexp, we cannot revoke the license key they memorized or copied. They can use our paid slots indefinitely on personal machines, and we are forced to play whack-a-mole removing them from the dashboard.
Conclusion:
If a product is marketed as having "10 seats," it must provide basic seat management (like individual user keys or key rotation). Selling a single, un-killable shared string and calling it "seats" is highly misleading.
I am keeping this at 2 stars for now. The functionality is easily worth 5 stars, but I cannot give it a top rating while my paid business assets are left entirely unprotected. I will gladly update this to 5 stars the moment true key rotation or per-user revocation is implemented.
------------------------------
Read less
Founder Team
Nicola_Vexp
Sep 20, 2026
Thank you for the honest review, and for the kind words about the product itself. Let me address the security framing carefully, because I think the category is wrong in a way that matters for anyone reading this.
A vexp license key does not grant access to your code, your repositories, your account, or any data of yours. vexp runs entirely on your machine with zero network calls. If someone obtained a key, what they'd gain is the ability to run vexp on their own computer, against their own code. That's lost revenue for me. Nothing of yours is exposed by it, which is the distinction between a licensing gap and a security breach.
On the part that is real: there's no key rotation, and removing a device frees the slot without invalidating the key. I've never claimed otherwise, and it's a genuine limitation. It comes directly from being local-first. There's no server between your machine and the license to revoke against, which is the same property that means your code never leaves your laptop and vexp keeps working offline. That tradeoff is deliberate, but it is a tradeoff.
What does exist today: device slots are finite (10 on Tier 4, not unlimited), every active device is visible at vexp.dev/account/devices, and you can remove any of them. For teams that need real per-person revocation now, individual licenses do that properly, since revoking one person actually means something when they hold their own key.
And on your scenario specifically, if a machine is compromised to the point where a license key is taken, the code itself, SSH keys, browser sessions and credentials all went with it. A rotation feature wouldn't have protected any of that. I say this not to dismiss the experience, which sounds genuinely awful, but because a license key is far down the list of what matters in that situation.
Per-user identity with proper revocation is the most requested thing I've had from Tier 3 and Tier 4 customers, and it's a direction I'm taking seriously. I won't promise a date I can't hold. Thank you for pushing on it, and I'm glad the rest of vexp is working well for you.
Helpful?
3
Share Vexp
Verified purchaser
Deals bought: 46Member since: Dec 2020
Aug 27, 2026
Come on, amazing tool for vibe coding (Trae)
Vexp is an amazing tool. It's super powers, YES, super powers make it a more than great companion for my vibe coding sessions using Trae. I'm on their ultra plan and I've spend more hours than ever vibe coding. It's neat. I'm at 62% saving with my "small" repo (4500 nodes). Vexp is everywhere I need it.
What surprised me is that I can now be "cooler" with my prompts. I don't need to overload Trae providing files or folders. Just using French language, yes, it perfectly reduces context and lowerizes token consumptons.
Vexp is a gem ! Tons of tacos 🌮
Read less
Founder Team
Nicola_Vexp
Aug 27, 2026
Thank you, this made our day. And you’ve put your finger on a benefit people rarely mention: not having to manually feed Trae files and folders anymore. Most of the attention goes to the token numbers, but not having to babysit context is often the bigger day-to-day win.
62% on a 4,500-node repo is solid, and it usually climbs as the project grows, since there’s more irrelevant code to leave out.
Also good to hear it holds up with French prompts. It resolves by identifiers and paths first, so naming real symbols helps in any language, but the surrounding prose being French clearly isn’t getting in the way.
Really glad it’s working for you, and enjoy those longer sessions. We’re here whenever you need us.
Helpful?
4
Share Vexp
Verified purchaser
Deals bought: 21Member since: Oct 2022
Aug 23, 2026
Una forma muy inteligente de dar contexto a los agentes de IA sin desperdiciar tokens
Llevo utilizando VEXP integrado con Visual Studio Code y Codex en varios proyectos de desarrollo reales y, hasta ahora, la experiencia está siendo muy buena.
Lo que más valoro es que no se limita a indexar el repositorio, sino que permite al agente obtener rápidamente el contexto relevante y entender las relaciones entre archivos, funciones, dependencias y tests sin tener que recorrer manualmente buena parte del código.
He hecho algunas pruebas específicas para comprobar hasta qué punto realmente aporta valor. Por ejemplo, en uno de mis proyectos pedí analizar el impacto de modificar una función concreta de autenticación. Con una única llamada a VEXP, el agente identificó la función, su caller directo, los componentes dependientes a varios niveles, el test relacionado y la documentación y ADR afectados. Todo ello sin realizar búsquedas adicionales ni abrir manualmente archivos del repositorio.
En ese tipo de tareas —análisis de impacto, arquitectura, dependencias o preparación de cambios que afectan a varios componentes— es donde estoy viendo más claramente el ahorro de contexto, tiempo y tokens.
También he comprobado que funciona mucho mejor cuando la consulta se ancla a archivos, funciones o símbolos concretos. En consultas demasiado genéricas puede recuperar documentación relevante pero no necesariamente todo el código afectado. No lo veo como un sustituto absoluto de las búsquedas tradicionales, sino como una capa adicional de contexto que permite que el agente empiece a trabajar con un mapa mucho mejor del proyecto.
La integración con Codex mediante MCP y los hooks automáticos resulta especialmente interesante: una vez configurado correctamente, el agente puede utilizar el índice del proyecto como parte natural de su flujo de trabajo.
Todavía estoy empezando a utilizarlo y quiero comprobar el ahorro a más largo plazo en repositorios mayores, pero las primeras pruebas han sido suficientemente buenas como para incorporarlo a mi forma habitual de trabajar con agentes de programación.
Además, se nota que es un producto todavía joven y en evolución, pero la idea que hay detrás me parece muy acertada: dar al agente el contexto que necesita, en lugar de hacerle consumir tokens descubriendo una y otra vez cómo está organizado el repositorio.
Muy buena compra hasta el momento.
Read less
Founder Team
Nicola_Vexp
Aug 23, 2026
The test you ran is exactly where the dependency graph earns its keep: identifying in a single call the function, its direct caller, dependents at several levels, the associated test, plus the documentation and ADR affected. That last part is one we particularly like, because documentation drifting out of sync with the code is a problem text search cannot solve by construction, and a graph that indexes markdown can.
And you're right about anchoring queries to specific files, functions or symbols. vexp resolves by identifier and path first, so an anchored question performs noticeably better than one in prose. Your framing matches ours too: it isn't a replacement for traditional search, it gives your agent a good map of the project before it starts. That's exactly the workflow we recommend.
Larger repositories are usually where the difference shows most, so we'd genuinely like to hear how it goes for you over time. Thanks for the purchase and for such a detailed analysis, we're here whenever you need us.
Helpful?
2
Share Vexp
Squibler PLUS
Verified purchaser
Deals bought: 18Member since: Feb 2025
Aug 17, 2026
Solid product with great support
While I wish that this tool had a GUI to make getting set up easier or to manage things like linked repositories visually or handle automatic updates; I was pleasantly surprised to see that AI tools I asked were able to figure it out how to set it up and configure vexp themselves pretty painlessly.
And since I do WordPress development, it's great being able to drag a plug-in into my dev folder and ask AI to connect it as a repository and then check some feature in that plugin so that I can work on an integration and AI will link it all together for me and then use Vexp's tools to locate the relevant code without needing to waste tons of time and tokens reading through files hunting for the starting point to my questions.
All in all it seems like a solid product that's getting regular updates and the customer support the few times I've reached out has been phenomenal. Looking forward to seeing how this tool continues to grow in the years to come!
Read less
Founder Team
Nicola_Vexp
Aug 18, 2026
Thank you, and the WordPress workflow you described is a great example of multi-repo doing its job: dropping a plugin in, connecting it as a repo, and having your agent trace the integration across both without reading through files hunting for a starting point. That's exactly the case it was built for.
On visual management and setup, that's fair feedback, and making those things easier to handle is something we're actively working on. Nice to hear your AI tools were able to configure it themselves in the meantime, that's become a surprisingly common path.
Really glad the support has landed well, and thanks for the kind words about the updates. We're here whenever you need us, and there's plenty more coming.
Helpful?
1
Share Vexp
Company details
Founder
\ See Vexp](https://vexp.dev/)
Founded February 26, 2026
🇮🇹Milan, Italy
1-10
Growth
Bootstrapped
Showing 1-1 of 1 posts
Updated Jun 23, 2026
Superpowers for your AI coding agent
I'm Nicola, and I built vexp because I was tired of watching my AI coding tools waste money reading code they didn't need.
Every time you ask Claude Code, Codex, or Cursor to fix a bug, the agent spends 15-20 tool calls just reading your project files before it writes a single line of code. That's 60% of your quota burned on orientation, not output. You're paying for the smartest models on the planet to do something a local index could do for free.
I built vexp to fix this. It's a context engine that runs entirely on your machine, maps your project's dependency graph, and hands your AI agent only the code that matters for each task. One call in...
Read more