How to Evaluate, Test, and Maintain FiveM Scripts

This no-fluff guide how to maintain FiveM scripts is for server owners, developers, and QA leads. You’ll get a production-like “Test City” in Docker, an acceptance checklist you can run end-to-end, a quantitative risk scoring model, and a vendor-vetting rubric that keeps you away from headaches.

This guide is part of our comprehensive FiveM scripts resource, where you'll find all our script recommendations, framework comparisons, and buying guides.

TL;DR

Spin up Test City (Docker) to isolate and benchmark any script safely.
Run the Acceptance Checklist before a penny changes hands.
Use the Risk Score (0–100) to decide ship/hold/reject.
Vet sellers with the Vendor Rubric (don’t skip this).
If you buy, prefer reputable stores — see our picks: Best Tebex Shops.

Part 1 — “Test City” (Docker) for Safe, Repeatable Script QA

What you get

Containerized FXServer + MariaDB (+ Adminer)
Clean server.cfg with oxmysql and base resources
Bind-mounted resources/custom where you drop the script under test
Deterministic network names/ports for simple DB strings

Download test-city.zip (Github)

How to Use Docker (FiveM)

How to use:

Unzip → cd test-city
Copy .env.example to .env and set your LICENSE_KEY (and DB creds if you like).
Drop oxmysql into server-data/resources/[standalone]/oxmysql/.
Put the script under test into server-data/resources/custom/<scriptname>/ and add ensure <scriptname> to server.cfg.
docker compose build && docker compose up -d → connect via Direct Connect to localhost:30120.

Prereqs: Docker + Docker Compose, a cfx license key, and an oxmysql resource copy (drop into server-data/resources/[standalone]/oxmysql).

Folder layout

text

test-city/
├─ docker-compose.yml
├─ fxserver/
│  ├─ Dockerfile
│  └─ entrypoint.sh
├─ server-data/
│  ├─ server.cfg
│  └─ resources/
│     ├─ [standalone]/oxmysql/   # place oxmysql here
│     └─ custom/                 # put the script under test here (e.g., myscript/)
└─ .env

`.env` (example)

text

LICENSE_KEY=changeme_cfx_license_key
MYSQL_DATABASE=fivem
MYSQL_USER=fivem
MYSQL_PASSWORD=fivempw
MYSQL_ROOT_PASSWORD=rootpw
FX_ARTIFACT_URL=https://runtime.fivem.net/artifacts/fivem/build_proot_linux/master/LATEST.tar.xz
# Tip: replace with a specific artifact URL you trust for reproducibility.

`docker-compose.yml`

text

version: "3.9"

networks:
  testcity:

volumes:
  db_data:

services:
  db:
    image: mariadb:10.11
    restart: unless-stopped
    environment:
      MYSQL_DATABASE: ${MYSQL_DATABASE}
      MYSQL_USER: ${MYSQL_USER}
      MYSQL_PASSWORD: ${MYSQL_PASSWORD}
      MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PASSWORD}
    command: >
      --character-set-server=utf8mb4 --collation-server=utf8mb4_unicode_ci
      --innodb_buffer_pool_size=256M
    volumes:
      - db_data:/var/lib/mysql
    networks: [testcity]

  adminer:
    image: adminer:4
    restart: unless-stopped
    ports:
      - "8080:8080"
    networks: [testcity]
    depends_on: [db]

  fxserver:
    build:
      context: ./fxserver
      args:
        FX_ARTIFACT_URL: ${FX_ARTIFACT_URL}
    restart: unless-stopped
    environment:
      LICENSE_KEY: ${LICENSE_KEY}
      MYSQL_DATABASE: ${MYSQL_DATABASE}
      MYSQL_USER: ${MYSQL_USER}
      MYSQL_PASSWORD: ${MYSQL_PASSWORD}
    depends_on: [db]
    networks: [testcity]
    ports:
      - "30120:30120/udp"
      - "30120:30120/tcp"
      - "40120:40120/tcp" # txAdmin (optional)
    volumes:
      - ./server-data:/opt/fivem/server-data

`fxserver/Dockerfile`

text

FROM debian:bookworm-slim

ARG FX_ARTIFACT_URL
ENV DEBIAN_FRONTEND=noninteractive

RUN apt-get update && apt-get install -y --no-install-recommends \
    xz-utils curl ca-certificates tini bash && \
    rm -rf /var/lib/apt/lists/*

# Fetch & extract FXServer artifact (URL provided via build-arg)
RUN mkdir -p /opt/fivem && \
    curl -fsSL "$FX_ARTIFACT_URL" -o /opt/fivem/fx.tar.xz && \
    tar -xJf /opt/fivem/fx.tar.xz -C /opt/fivem && \
    rm -f /opt/fivem/fx.tar.xz

# Non-root
RUN useradd -ms /bin/bash fivem
USER fivem

WORKDIR /opt/fivem
COPY --chown=fivem:fivem ../server-data /opt/fivem/server-data
COPY --chown=fivem:fivem entrypoint.sh /entrypoint.sh

ENTRYPOINT ["/usr/bin/tini","--"]
CMD ["/bin/bash","/entrypoint.sh"]

`fxserver/entrypoint.sh`

text

#!/usr/bin/env bash
set -euo pipefail

# Provide DB string for oxmysql via server.cfg (already set there),
# just ensure the DB is reachable before boot to avoid spam errors.
echo "Waiting for database..."
until nc -z db 3306; do sleep 1; done
echo "DB ready."

# Run FXServer with our server.cfg
# Most Linux artifacts ship a run.sh in the artifact root.
exec bash /opt/fivem/run.sh +exec /opt/fivem/server-data/server.cfg

If your artifact uses a different launcher path, adjust the last line accordingly (e.g., /opt/fivem/alpine/opt/cfx-server/run.sh).

`server-data/server.cfg` (lean, test-ready)

text

# ---------- Identity ----------
sv_licenseKey "%LICENSE_KEY%"            # injected by env in entrypoint command or replace manually
sv_hostname "Test City — Script QA"
sets tags "qa,testing,fivemx"
sv_maxclients 2
onesync on
sv_scriptHookAllowed 0
sv_enforceGameBuild 3095

# ---------- Networking ----------
endpoint_add_tcp "0.0.0.0:30120"
endpoint_add_udp "0.0.0.0:30120"

# ---------- Database (oxmysql) ----------
set mysql_connection_string "mysql://%MYSQL_USER%:%MYSQL_PASSWORD%@db:3306/%MYSQL_DATABASE%?charset=utf8mb4"

# ---------- Core / Base resources ----------
ensure mapmanager
ensure chat
ensure spawnmanager
ensure sessionmanager
ensure hardcap
ensure baseevents

# ---------- Datastore ----------
ensure oxmysql

# ---------- Under test ----------
# Drop your script folder into resources/custom/<scriptname> and enable here:
# ensure myscript

# ---------- QA helpers ----------
# Verbose logging in dev
setr con_minSeverity info

# Enable txAdmin panel (optional)
setr txAdminPort 40120

Enable your script: copy it into server-data/resources/custom/myscript/ and add ensure myscript to the bottom block.

Run it

text

cd test-city
docker compose build
docker compose up -d
# Adminer at http://localhost:8080 (server: db, user: fivem, pass: fivempw)
# Connect FiveM client → Direct Connect: your-docker-host:30120

Part 2 — Script Acceptance Checklist (Do Not Skip)

Use this every time. Copy into your tracker and check off items.

A. Pre-flight (before install)

Source type: Open source / partially open / escrow-only.
Dependencies listed: framework (ESX/QBCore/QBOX), ox_lib, oxmysql, PolyZone, etc.
fxmanifest.lua: lua54 'yes', correct game 'gta5', fx_version not ancient.
Docs: install steps, config examples, permissions, locales, known conflicts.
License/ToS: usage rights, refund policy, updates, support channel.

B. Installability

No console errors on ensure: zero stack traces or asset missing spam.
No deprecated natives spam.
Config loads clean: no JSON/Lua syntax errors.
DB migrations: create tables once; re-starts don’t re-apply or break.

C. Functional

Core flows work: happy path tested (e.g., purchase, job flow, crafting).
Edge cases: wrong inputs, missing perms, out-of-range quantities.
Localization: no hard-coded strings if locales promised.
Permissions: staff/admin commands gated; no free access client-side.

D. Performance (client & server)

resmon: idle ≤ 0.10ms, active ≤ 0.50ms (client).
Server tick health: no long hitch warnings under normal use.
DB queries: no tight loops; prepared statements used; minimal N+1.

E. Security

Server-side validation: every server event validates source, perms, and input types.
No trust in client: money/inventory changes only from server logic.
No eval/loadstring/remote code patterns.
Anti-abuse: rate limits/throttles on spammy endpoints.
Escrow integrity: no backdoors in non-escrow config loaders.

Good event pattern (example):

text

RegisterNetEvent('myscript:buy', function(item, amount)
  local src = source
  if type(item) ~= 'string' or type(amount) ~= 'number' or amount < 1 then return end
  if not HasPermission(src, 'shop.buy') then return end
  local xPlayer = GetPlayer(src) -- framework adapter
  if not xPlayer then return end
  -- validate price server-side, check inventory limits, do DB in a transaction
end)

F. Compatibility & Cleanliness

Framework adapters present (ESX/QBCore/QBOX) or clearly declared unsupported.
No globals leakage, unique event names, no resource name assumptions.
Uninstall clean: removing the resource doesn’t break others.

Part 3 — Quantitative Risk Scoring Model (0–100)

Use this to decide “ship/hold/reject.” Lower is better.

Factor	Weight	How to Score (0=good → 5=bad)
Performance	0.20	resmon idle: 0=≤0.10ms, 1=≤0.20, 3=≤0.50, 5=>0.50; spikes add +1
Security	0.25	0=all server-validated + rate limits; 3=some gaps; 5=trusts client / unsafe events
Stability	0.15	0=no errors; 3=occasional warnings; 5=frequent errors/hitches
Compatibility	0.10	0=adapters/tests; 3=one framework only; 5=breaks common deps
Maintainability	0.10	0=docs/changelog/semantic versions; 5=no docs/abandoned
Supply-chain/Vendor	0.20	0=reputable, history, refunds; 5=unknown, no policy

Formula

text

RiskScore = 100 * Σ(weight_i * (score_i / 5))

Bands & Actions

0–24 (Low): Ship to staging, monitor.
25–49 (Moderate): Fixes required before live.
50–74 (High): Hold; request vendor patches or replace.
75–100 (Critical): Reject.

Example

Perf=1, Sec=3, Stability=1, Compat=1, Maint=2, Vendor=1
Risk = 100*(.2*.2 + .25*.6 + .15*.2 + .1*.2 + .1*.4 + .2*.2) = 34 → Moderate

Part 4 — Vendor Vetting Rubric (score each 0–5, higher is better)

Criterion	What “Good” Looks Like	Notes
Identity & Track Record	Clear brand, years active, consistent handle	Avoid burner stores
Documentation	Install + config + perms + troubleshooting	Screenshots/gifs help
Update Cadence	Changelog, semantic versions, recent commits/releases	Quarterly+ is fine
Support Quality	Ticket/Discord SLAs, reproducible fixes, not just “restart”	Sample responses
Issue Transparency	Public known issues & roadmap	Honesty > perfection
Refund/ToS Clarity	Refund window, conditions, license terms	No dark patterns
Testing Proof	Staging video, perf metrics, framework list	Even better: demo server
Security Hygiene	Mentions server-side checks, no sensitive keys, no eval	Ask about audits
Compatibility Policy	ESX/QBCore/QBOX stated, adapters, version matrix	State supported builds
Pricing & Value	Fair for complexity, no dependency paywalls	Watch upsell traps

Rubric Score (0–50):

40–50: Strong — preferred vendor
30–39: Acceptable — monitor
20–29: Weak — proceed cautiously
<20: Avoid

Pro tip: Cross-reference with reputable marketplaces. Start here: Best Tebex Shops.

Part 5 — Maintenance Playbook (After You Ship)

Version pinning: lock script releases + dependency versions (e.g., oxmysql, ox_lib).
Staging first: all updates land in Test City; run the acceptance checklist again.
Backups & rollback: DB snapshot + resource files before each update. Keep a Rollback.md with exact steps.
CI smoke test (optional, recommended): headless client connect + command macro to hit main flows; parse server console for errors.
Operational metrics: keep a simple ledger per script: average resmon (idle/active), error count/session, incident notes.
Changelog discipline: require vendor changelogs; maintain your own integration notes.
De-risk scheduled events: update outside peak hours; announce maintenance windows.

Part 6 — Templates You Can Copy

A. Test Plan (per script)

text

# Script Test Plan — <name> <version>

## Context
Framework(s): ESX/QBCore/QBOX
Dependencies: oxmysql vX, ox_lib vY, PolyZone vZ
DB Migrations: yes/no
Escrow: yes/no

## Functional Cases
- [ ] Case 1:
- [ ] Case 2:
- [ ] Negative 1:

## Performance
resmon idle: ____ ms
resmon active: ____ ms (scenario: ______)

## Security Checks
- [ ] All server events validate source/perm/input
- [ ] Rate limits present
- [ ] No client-trusted money/inventory mutations

## Logs & Errors
Paste snippet (server + F8): 

## Result
Pass/Fail + Notes

B. Rollback.md (skeleton)

text

# Rollback — <date> <script> <from→to>

1) Disable script: `stop <resource>`
2) Restore resource files from backup: <path>
3) Restore DB snapshot: <path/command>
4) Restart FXServer
5) Verify: console clean, perf normal, flows ok

C. Issue Template

text

**Summary**
What happened vs expected, with timestamps.

**Repro Steps**
1) ...
2) ...

**Env**
Artifact build:
Framework & versions:
Script version:
Other deps:

**Logs**
Server console:
F8:

**Attachments**
Screens/video if possible.

How to Use This Guide Efficiently

Bootstrap Test City once, keep it clean.
For each candidate script:
- Drop into resources/custom/, enable in server.cfg.
- Run the Acceptance Checklist.
- Compute the Risk Score.
- Vet the seller with the Vendor Rubric.
- If it passes, merge into staging; if not, request fixes or walk away.

Need more testing guides? Check out the section

TL;DR

Part 1 — “Test City” (Docker) for Safe, Repeatable Script QA

Folder layout

.env (example)

docker-compose.yml

fxserver/Dockerfile

fxserver/entrypoint.sh

server-data/server.cfg (lean, test-ready)

Run it

Part 2 — Script Acceptance Checklist (Do Not Skip)

A. Pre-flight (before install)

B. Installability

C. Functional

D. Performance (client & server)

E. Security

F. Compatibility & Cleanliness

Part 3 — Quantitative Risk Scoring Model (0–100)

Part 4 — Vendor Vetting Rubric (score each 0–5, higher is better)

Part 5 — Maintenance Playbook (After You Ship)

Part 6 — Templates You Can Copy

A. Test Plan (per script)

B. Rollback.md (skeleton)

C. Issue Template

How to Use This Guide Efficiently

Mantente al Día

`.env` (example)

`docker-compose.yml`

`fxserver/Dockerfile`

`fxserver/entrypoint.sh`

`server-data/server.cfg` (lean, test-ready)