Last updated

Vibe-Coded App Rescue

Fix My Claude Code App

From
£1,000
Deliverables
6
Process
5 steps
Studio
Telford, UK

Built your app with Claude Code and it is not quite production-ready? We review what the agent wrote, fix the root causes with real tests, lock down secrets and permissions, and leave a Claude Code setup you can keep building with safely.

In short

Circlemount, a Telford, Shropshire software team, fixes apps built with Claude Code for UK founders and businesses. It reviews the code and agent setup, fixes root causes with real tests, removes exposed secrets, makes deployment reliable and leaves a safe Claude Code configuration. The initial app assessment is free; projects start from £1,000 + VAT.

5.0 Google (17 reviews)£57m client revenue213 clients

The Study Room website, designed and built by Circlemount
The Study Room  ·  Private tuitionCase study →
01The problem

Claude Code writes a lot of code, fast. Somebody still has to own it.

Claude Code is Anthropic's agentic coding tool: it reads your codebase, edits files, runs commands and can work through a whole feature from one instruction. Used well, it is remarkably productive. We use it ourselves, alongside Antigravity, Codex and Lovable, to build our own products.

The trouble starts when nobody reviews what it does. The agent reports a task as finished when the tests never ran. A long session is compacted and forgets how the app deploys. A command approved with “don't ask again” is saved, secret and all, into a settings file that later reaches a public repository: Apiiro found 1,379 secrets in public copies of that file in November 2026. Each change is plausible on its own, and the codebase drifts into something nobody fully understands.

We review what Claude Code produced, find the real causes, fix them with tests and proper deployment, and set Claude Code up so it keeps helping: clear project instructions, tight permissions, hooks that block risky actions, and review before anything reaches production.

02Who it's for

Sound familiar?

  • 01

    “Claude says it's done, but it isn't”

    Features reported as finished that fail when real users try them, and tests that never ran.

  • 02

    “It ran a destructive command”

    Files, folders or data deleted or overwritten during a session, and no clear record of what else changed.

  • 03

    “It forgot how we deploy”

    Long sessions that lose the thread after compaction, and builds that only work on one machine.

  • 04

    “Our keys ended up in .claude”

    Secrets typed into approved commands, saved in settings files and copied where they should not be.

  • 05

    “Nobody understands the codebase any more”

    Thousands of lines written faster than anyone could review them, and fear of every new change.

03Deliverables

What We Deliver

  • 01

    Code and agent setup review

    A plain-English review of the code Claude Code wrote, the git history and the agent settings: what is solid, fragile or unsafe.

  • 02

    Root-cause fixes with real tests

    Fixes for the bugs that keep coming back, with tests that exercise real behaviour, so “tests pass” means something.

  • 03

    Secrets and permissions

    Secrets moved out of code and agent settings, exposed keys rotated, and .claude files kept out of the repository.

  • 04

    Deployment that works

    Repeatable builds and environments on Vercel, Cloudflare, Netlify or your platform, with staging and rollback.

  • 05

    A safe Claude Code setup

    CLAUDE.md project instructions, allow and deny rules, hooks and review steps your team can keep using.

  • 06

    Security, speed and SEO

    Database access rules, authentication, performance and search visibility checked and fixed.

04Process

How It Works

  1. 1. Share the repository

    Read access to the code and a description of what is broken or blocking launch.

  2. 2. Free app assessment

    We review the code, history, settings and deployment, and reproduce the problems.

  3. 3. Proposal

    What we will fix and in what order, the timeline and a fixed price.

  4. 4. Fix, test, deploy

    Work in branches with tests, a staging copy for you to check, then production.

  5. 5. Handover or support

    A walkthrough of the new setup, or an ongoing support plan.

05How we harden it

From agent output to production

How we make a Claude Code app production-ready
  1. AssessRead the code, git history and .claude settings; reproduce what is broken.
  2. ContainRotate exposed secrets, tighten allow and deny rules, separate production access.
  3. ProveTests that fail before the fix and pass after it, against real behaviour.
  4. ShipStaging first, then a repeatable production deploy with rollback.
  5. GuardCLAUDE.md, hooks and a review step so the next session builds on solid ground.

Sources: Claude Code settings; Claude Code permission modes; The Hacker News on Apiiro's findings.

06Related

Rescuing an app, or rolling Claude Code out?

This page is for apps Claude Code has already built. To build agents with Claude, or to set Claude Code up across a team, start here instead.

07Pricing

What it costs

The initial app assessment is free, project work starts from £1,000 + VAT, and ongoing support plans start from £125 a month + VAT.

  • 01

    Free app assessment

    We review the code, git history and Claude Code settings, then send a written, tailored proposal. No obligation.

  • 02

    Projects from £1,000 + VAT

    Fixes, features, security and launch work, priced for your app in a written proposal before anything starts.

  • 03

    Support from £125 a month + VAT

    Monitoring, updates, fixes and help as you keep building, on a monthly plan you can stop.

08Our work

Apps we've built with AI, running in production

We build with AI coding agents every day, so we know the habits they bring to a codebase.

The Study Room's Portal, Dentorb and ORBQ were built with AI-assisted development, using Antigravity, Claude Code, Codex and Lovable, and every change was reviewed, tested and owned by an experienced developer. We know where these tools are strong, where they cut corners, and what production asks of the code they write.

11Where we work

We Serve Businesses Across the UK

Based in Telford, we provide Fix My Claude Code App to businesses nationally including:

12FAQs

Frequently Asked Questions

Can you fix an app built with Claude Code?

Yes. Circlemount fixes apps written with Claude Code for UK founders and businesses. We review the code, the git history and the agent settings, fix root causes with tests that check real behaviour, clean up secrets and permissions, make deployment reliable, and leave a Claude Code setup your team can keep using safely.

Why does Claude Code say a task is done when it isn't?

The agent judges its own work from what it can see in the session. If tests were skipped, mocked or never written, or the context was compacted mid-task, it can report success that the code does not support. Real tests, review gates and hooks that run checks make “done” verifiable.

Claude Code deleted files or ran a destructive command. Can you help?

Yes. First we stop further damage, recover what version control, backups or your host can restore, and check what else the session changed. Then we fix the cause: deny rules for destructive commands, separate credentials for production, hooks that block risky actions, and a review step before anything touches live data.

Is my Claude Code project leaking secrets?

Check now. Claude Code saves approved commands in .claude/settings.local.json and keeps that file out of git by default, but copies still reach public repositories, and any secret typed into an approved command is stored with it. We search the code, history and settings, rotate exposed keys and move secrets server-side.

Can I keep using Claude Code after you fix my app?

Yes, that is usually the goal. We leave a CLAUDE.md that explains the project, permission rules that match your risk, hooks that run tests and block dangerous commands, and a review routine. Your team keeps Claude Code's speed while the codebase stays tested and understood.

Do you use Claude Code yourselves?

Yes. Our developers use Claude Code every day, alongside Antigravity, Codex and Lovable, to build products such as The Study Room's Portal, Dentorb and ORBQ. Every change is reviewed, tested and owned by an experienced developer. Circlemount is independent and not an Anthropic partner.

How much does it cost to fix a Claude Code app?

The initial app assessment is free: we review the code, history and agent settings and send a tailored proposal. Project work starts from £1,000 + VAT, depending on what is broken and how large the app is, and ongoing support plans start from £125 a month + VAT.