Speed — does it scale?
Brief 05

10x Stress Review

Simulate 10x users, data, and traffic on paper — find what breaks first, at what threshold, and the cheapest mitigation.

SCALE.md4 phases7 lenses~2.9k chars

Free & open · no signup · read-only — it ends by asking · nothing leaves your machine

What it does

You are working inside this repo. Mission: put tomorrow's load on today's architecture — 10x the users, 10x the data, 10x the traffic — and find what breaks first, in what order, and what each fix costs.

This is a thought experiment on tomorrow's load, the whole product included — cost, ops, third parties. For today's configured ceilings, measured from config, run 123.

Read-only pass, thinking on paper. Your only write is the report file.

What it looks for

7 lenses it looks through

Phase 2 sweeps the codebase through every one of these, citing file and line for each finding.

1

10x data

unpaginated queries, full scans, unbounded UI lists, tables with no archival story

2

10x users

connection pools, session storage, rate limits, lock contention, auth bottlenecks

3

Single points of failure

one-instance assumptions, local file state, in-memory queues/caches that don't survive restarts or replicas

4

Jobs & queues

backlog behavior under load, retry storms, poison messages, no dead-letter path

5

Third-party ceilings

API quotas and rate limits you'll hit at 10x

6

Cost curve

what scales linearly vs what explodes (per-request third-party calls, storage egress)

7

Operational load

manual steps that become 10x painful (deploys, support lookups, data fixes)

The deliverable

What lands in your repo

One structured report at the repo root — or in reports/, if you keep one — the same shape every time, ready for a teammate — or the optional Studio — to act on.

SCALE.md
1
Today's scale snapshot
the numbers and assumptions
2
First-to-break ranking
each: bottleneck · breaking mechanism · estimated threshold · symptom users would see
3
Mitigations
cheap patch and real fix per bottleneck, with effort
4
Do-not-fix-yet list
premature optimizations to consciously skip
5
The one change
biggest headroom per unit of effort
One file. Evidence-backed. It ends by asking before touching anything.
How it works

What it makes the agent do

Every Goal Prompt follows the same four steps, so results are consistent and repeatable — no matter which one you run.

1
Phase 1

Establish today's scale

Infer current scale from code and config.

2
Phase 2

Stress through 7 lenses

For each, name the breaking mechanism and estimate the threshold.

3
Phase 3

Curate

Order findings by which breaks first as load grows.

4
Phase 4

Report

Create SCALE.md at repo root.

Get started

Three ways to run this Goal Prompt

01 · COPY

Paste it in

Copy it and paste it into your agent inside the repo you want checked.

02 · INSTALL

As a slash command

Install the goal plugin once — two commands — then just type /goal:stress-review-10x.

1/plugin marketplace add GhostlyGawd/goal-prompts
2/plugin install goal@goal-prompts

Or install only this Goal Prompt as /goal-stress-review-10x:

curl -fsSL https://goal-prompts.vercel.app/install | BRIEF=05 sh
03 · AGENT

From an agent (MCP)

Let an agent fetch it mid-conversation, or pull the raw Goal Prompt by URL.

https://goal-prompts.vercel.app/raw/05.md
Transparency

The exact prompt

Nothing hidden — this is the whole Goal Prompt, verbatim. Read it in a minute, edit it, or copy it as-is.

Read the full brief (2,883 characters)
# Goal: 10x Stress Review

You are working inside this repo. Mission: put tomorrow's load on today's architecture — 10x the users, 10x the data, 10x the traffic — and find what breaks first, in what order, and what each fix costs.

This is a thought experiment on tomorrow's load, the whole product included — cost, ops, third parties. For today's configured ceilings, measured from config, run 123.

Read-only pass, thinking on paper. Your only write is the report file.

## Phase 1 — Establish today's scale
- Infer current scale from code and config: pool sizes, rate limits, page sizes, batch sizes, timeouts.
- Which growth axis matters most for this product: users, data volume, traffic spikes, or tenant count?
- Which tables, queues, and stores grow without bound?

## Phase 2 — Stress through 7 lenses
For each, name the breaking mechanism and estimate the threshold.
1. **10x data** — unpaginated queries, full scans, unbounded UI lists, tables with no archival story
2. **10x users** — connection pools, session storage, rate limits, lock contention, auth bottlenecks
3. **Single points of failure** — one-instance assumptions, local file state, in-memory queues/caches that don't survive restarts or replicas
4. **Jobs & queues** — backlog behavior under load, retry storms, poison messages, no dead-letter path
5. **Third-party ceilings** — API quotas and rate limits you'll hit at 10x
6. **Cost curve** — what scales linearly vs what explodes (per-request third-party calls, storage egress)
7. **Operational load** — manual steps that become 10x painful (deploys, support lookups, data fixes)

## Phase 3 — Curate
- Order findings by which breaks first as load grows
- For each: cheap patch vs real fix — name both
- Explicitly list what NOT to fix yet (premature scaling is also a bug)

## Phase 4 — Report
Create `SCALE.md` at repo root:
1. **Today's scale snapshot** — the numbers and assumptions
2. **First-to-break ranking** — each: bottleneck · breaking mechanism · estimated threshold · symptom users would see
3. **Mitigations** — cheap patch and real fix per bottleneck, with effort
4. **Do-not-fix-yet list** — premature optimizations to consciously skip
5. **The one change** — biggest headroom per unit of effort

Start the report with today's date. If `SCALE.md` already exists from a previous run, read it first and lead with what changed since.

## Rules
- Every bottleneck needs a mechanism — "might be slow" doesn't count
- Respect today's simplicity: recommend the least architecture that survives 10x
- No load-bearing system in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
- If a `reports/` directory exists at the repo root, write the report there instead of the root.
- Before asking, present the top findings as a ranked list in plain words
- Report only — end by asking which mitigations to build

House rules for this Goal Prompt

  • Every bottleneck needs a mechanism — "might be slow" doesn't count
  • Respect today's simplicity: recommend the least architecture that survives 10x
  • No load-bearing system in this repo? Say so in a one-paragraph null report and stop — a null result is a valid finding.
  • If a reports/ directory exists at the repo root, write the report there instead of the root.
  • Before asking, present the top findings as a ranked list in plain words
  • Report only — end by asking which mitigations to build
Keep exploring

Pairs well with

Curated neighbors — briefs that answer the adjacent question, worth running in the same session.

More Speed briefs