Turns a bug report into a reproduction that fails, before anybody writes a line of the fix.
Built for: Whoever picks up the report that says it happens sometimes, and has to decide within an hour whether it is real.
The button opens the checkout, where 6 servers and 5 skills are listed one by one with what each does to the bill — free, already yours, monthly or a one-off licence. Nothing is charged until you confirm it there, in Stripe’s own card frame on that page rather than a redirect, and each paid member keeps its own budget cap.
The reason to buy a stack rather than five listings: each line below needs more than one member connected at the same time.
Pull the error, its stack trace and its frequency out of Sentry, and say how many people it reached.
Replay the path in a real browser with Puppeteer and stop when the page does not fail the way the report says it does.
Read the rows behind the report against a read-only Postgres connection with row caps and a statement timeout.
Write the reproduction down as a markdown note with a diagram, so the next person starts where this one stopped.
What each part contributes, and why it was picked over the obvious alternative. Prices and permissions are read from the listings, so nothing here can disagree with the catalogue.
The report as the system recorded it: the event, the release and how often it happened, which is what decides whether this is worth an hour.
A local log server the app posts structured events to, with a live tail — the missing half when the stack trace stops at the boundary.
Read-only Postgres with row caps and statement timeouts, so checking the data behind a bug cannot become an incident of its own.
Drives a real browser through the reported path, which is the only way to tell a broken page from a misremembered one.
Views and edits the files the reproduction lives in — the failing script, the fixture, the note — without giving the agent a shell.
The official knowledge-graph server keeps what was already ruled out, so the second hour does not re-test the first hour's dead ends.
Holds the desk to root cause: reproduce, isolate, then explain — rather than the first plausible change that makes the symptom go away.
Screenshots, console and performance traces through Puppeteer scripts, so 'it is slow' turns into a number from the profile.
Researches the codebase and the schema before writing the end-to-end case, which is how a reproduction ends up in the suite instead of in a comment.
Writes the reproduction up with the sequence diagram attached, so the handover is a document rather than a paragraph of chat.
Finds the incident learning or workaround that explains the odd code, which is often the bug's actual origin.
Every member is a subscription or a licence bought once, so the monthly figure is a price rather than an estimate: what moves it is adding or dropping a member, not how hard the stack is worked. The one assumption is that a one-off licence is spread over a year so it can sit in the same column as a subscription.
| Part | What you are paying for | Monthly, as quoted |
|---|---|---|
| Sentry | Free | — |
| Logsafe | Free | — |
| Postgres | Free | — |
| Puppeteer | Free | — |
| Filesystem | Free | — |
| Memory (official) | Free | — |
| Skills | ||
| Debugging and Error Recovery | $19 · $1.58/mo over 12 months | $1.58/mo |
| Chrome DevTools | Free · context cost only | — |
| E2E Test | Free · context cost only | — |
| Markdown and Mermaid Writing | Free · context cost only | — |
| Keep the Why | $79 · $6.58/mo over 12 months | $6.58/mo |
| Everything above | $0 of servers plus $8.00 of skills, the same in a quiet month and a busy one | $8.00/mo |
Subscriptions at their monthly plan price; one-off licences spread over 12 months. One-off purchases in this stack total $98 — Debugging and Error Recovery $19, Keep the Why $79 — paid once and spread here so they sit in the same column as a subscription. Everything arrives on one mcprush invoice, taken by Stripe from the card on your account, not one per publisher — mcprush.com is the merchant of record and each publisher is paid out of it.
Install only these and the bill is nothing: 6 servers and 3 skills, 4 of the 4 tools.
Left out, and what goes with it:
Nothing in this stack carries a call allowance, so nothing here can run out before the month does. The bill is decided when you install it, not when you use it.
Budget caps are set per install and enforced at the gateway, so a retry loop is refused at the cap rather than left to run through an allowance overnight.
One step per part, in the order they are useful: connect what the work reads before what it writes, and install the skills that decide how the work is done last. Each step is a command you can read before you run it.
npx mcprush@latest stack add bug-reproNothing in this stack installs from one command today: 11 members are either paid, run from its own source, or a skill with its own command — the steps above name each one. Nothing is connected until you approve it.
Installed together, these tool surfaces add up. It is the first thing a security reviewer asks for, so what has been counted — and what nobody has counted yet — is on the page rather than in a PDF.
| Member | Tool surface | Write tools |
|---|---|---|
| Filesystem | not imported | not recorded |
| Memory (official) | not imported | not recorded |
| Postgres | not imported | not recorded |
| Puppeteer | not imported | not recorded |
| Sentry | not imported | not recorded |
A stack's blast radius is the union of its members, not the worst of them. That union cannot be taken here, because 5 members of 6 have no imported tool surface — so the figure a review asks for is missing rather than low, and a member marked not imported is one nobody has counted rather than one that cannot write.
1 skill in this stack instructs the agent to write — Markdown and Mermaid Writing. A skill holds no permission of its own; it can only ask a server that already has the scope, and the approval gate still applies.
A stack is a default, not a verdict. These are the substitutions the maintainer would make, and what each one costs or saves.
Where the flow needs an agent that decides its own next click instead of a script that repeats yours, take the browser-use server and give up the step-by-step replay.
The same read-only guardrails, row caps and statement timeouts against MySQL, for the half of the estate that never moved to Postgres.
Written by the maintainer, kept on the page rather than in a support thread.
It cannot fix the bug in your repository. Nothing here has write access to code — the desk hands over a reproduction, not a patch.
Postgres access is read-only by design, so a data-level bug can be found here but never corrected here.
The browser member is an experimental Puppeteer server. A flow behind a login wall or a CAPTCHA is where it stops, and the desk should say so rather than keep clicking.
A stack has an owner: whoever keeps it re-checks the combination when a member changes, and the members themselves are published by the people named on each row.
Reads the ticket, finds whether it is a bug or a billing question, and drafts the reply in your own voice.
Reads the branch, runs it in a throwaway sandbox, and leaves the review on the merge request before a human opens it.
Answers the page: what fired, what changed, what the runbook says, and what to tell the channel.