What is TjAgentForge?

TjAgentForge is an Agent-as-a-Service (AaaS) platform for mobile game studios. It connects attribution, ad revenue, IAP and event data sources, then runs an always-on AI operations employee that detects metric anomalies, suggests and applies parameter tuning, opens code fix pull requests and produces diagnosis reports.

Does the AI execute changes without approval?

No. By default every write action — deleting resources, changing members, rolling out tuning or opening a pull request — requires your confirmation. Lower-risk writes can be auto-executed by changing the automation level, while high-risk actions still need approval. Every change keeps a rollback point and an audit record.

Do I need to upload my source code?

No. Local repositories are handled by the CLI Agent running on your own machine; it only runs whitelisted tasks that you authorize per execution, so source code never leaves your computer.

How do I integrate TjAgentForge with my own systems?

Use the REST Open API or the MCP Server. Both share the same API key, execution path and audit trail; keys are scoped read / write / admin, and write calls return a confirmation link unless the tenant is set to full automation.

How is TjAgentForge priced?

Pricing is based on seats plus a monthly task quota. Signing up includes a free starter seat; paid tiers are Pro, Team and Enterprise, and extra seats or task quota can be added separately. Prices shown on the site are indicative and confirmed in the console.

What data can it connect to?

Four kinds of sources: attribution, ad revenue, in-app purchases (IAP) and event data. Once connected, the always-on AI operations employee watches them in one place — anomaly detection, tuning and reports — instead of you hopping between dashboards.

What if the AI makes a wrong change?

Every change leaves a rollback point and an audit record, so it can be reverted on demand; write operations require your confirmation by default. You can also lower the automation level at any time so more actions stay pending your approval.

Do I need to install anything?

Not for data — after signing up you connect data sources in the console. You install the CLI Agent on your own machine only when you want the AI to operate your local git repos (branch, apply patch, commit). It ships as a Python-free installer for macOS, Linux and Windows.

Which languages are supported?

The site and console are available in English, Simplified Chinese, Traditional Chinese, Japanese and Korean, switchable at any time.

Can I try it for free?

Yes. Signing up includes a free starter seat, so you can connect a data source and see anomaly detection and diagnosis reports. Upgrade to a paid tier when you need more seats or task quota.

TjAgentForge

Hire an AI Operator

It watches your metrics 24/7 and acts — tuning, fixing, reporting. High-risk actions wait for your approval, and every step can be rolled back and is audited.

Start free (first month free)See what it does

What it does for you every day

Not another dashboard — a teammate that actually does the work.

Watches your data

Connects attribution / ad revenue / IAP / analytics sources, sweeps metrics 24/7 and triages anomalies per project.

Tunes for you

An anomaly triggers AI diagnosis that names the exact knob and target value; apply it and roll back anytime.

Fixes code

Locates crashes, generates a diff and opens a PR; local repos go through CLI Agent — source never leaves your machine.

Writes reports

Ask anytime for a diagnosis report built on real metrics and anomalies, with clear next actions.

Counts time saved

Every AI action is converted into saved ops hours and summarized in a time-saved bill.

Touches your machine

When local access is needed, CLI Agent does it: allow-listed tasks only, and each run requires your authorization.

AI acts, you decide

High-risk actions need your nod

On the default level, deletions, member changes, tuning rollouts and PRs always show a confirm card first.

Every step can be rolled back

Write actions take a rollback snapshot first; a wrong call can be reverted in one click (non-reversible ones are flagged).

Fully auditable

Who asked the AI to do what, and how it turned out, is recorded in the audit log.

Pay per seat, task quota included

The tier sets how many seats can run in parallel and the monthly task quota; extras can be purchased.

TrialFree

Free

Seats1
Task quota / mo50
Local access (CLI Agent)—
Data accumulation—
Vertical model—

Start free

Standard

$149/mo

Seats1
Task quota / mo300
Local access (CLI Agent)✓
Data accumulation✓
Vertical model—

Start free

Team

$399/mo

Seats5
Task quota / mo1200
Local access (CLI Agent)✓
Data accumulation✓
Vertical model—

Start free

Enterprise

Custom

SeatsUnlimited
Task quota / moUnlimited
Local access (CLI Agent)✓
Data accumulation✓
Vertical model✓

Start free

The Trial seat is a free tier — just sign up, monthly task quota included. Paid tiers are priced for reference; the final quote is on the console's Seats & Task Quota page.

CLI Agent (local access)

Operate your local git repos without uploading source code. The Agent runs on your machine and only executes allow-listed tasks with your authorization.

Download

Download the standalone build for your platform — no Python required.

macOS (Apple Silicon)macOS (Intel)Linux (x86_64)Windows (x86_64)

Generate token

Create a connection token in the console.

Start & connect

Launch the Agent and confirm it appears online.

Install guide

Frequently asked questions

TjAgentForge is an Agent-as-a-Service (AaaS) platform that hires you an always-on AI operations employee for mobile games: it watches your data, acts on anomalies, fixes code and reports back — while humans keep the final decision.

What is TjAgentForge?: TjAgentForge is an Agent-as-a-Service (AaaS) platform for mobile game studios. It connects attribution, ad revenue, IAP and event data sources, then runs an always-on AI operations employee that detects metric anomalies, suggests and applies parameter tuning, opens code fix pull requests and produces diagnosis reports.
Does the AI execute changes without approval?: No. By default every write action — deleting resources, changing members, rolling out tuning or opening a pull request — requires your confirmation. Lower-risk writes can be auto-executed by changing the automation level, while high-risk actions still need approval. Every change keeps a rollback point and an audit record.
Do I need to upload my source code?: No. Local repositories are handled by the CLI Agent running on your own machine; it only runs whitelisted tasks that you authorize per execution, so source code never leaves your computer.
How do I integrate TjAgentForge with my own systems?: Use the REST Open API or the MCP Server. Both share the same API key, execution path and audit trail; keys are scoped read / write / admin, and write calls return a confirmation link unless the tenant is set to full automation.
How is TjAgentForge priced?: Pricing is based on seats plus a monthly task quota. Signing up includes a free starter seat; paid tiers are Pro, Team and Enterprise, and extra seats or task quota can be added separately. Prices shown on the site are indicative and confirmed in the console.
What data can it connect to?: Four kinds of sources: attribution, ad revenue, in-app purchases (IAP) and event data. Once connected, the always-on AI operations employee watches them in one place — anomaly detection, tuning and reports — instead of you hopping between dashboards.
What if the AI makes a wrong change?: Every change leaves a rollback point and an audit record, so it can be reverted on demand; write operations require your confirmation by default. You can also lower the automation level at any time so more actions stay pending your approval.
Do I need to install anything?: Not for data — after signing up you connect data sources in the console. You install the CLI Agent on your own machine only when you want the AI to operate your local git repos (branch, apply patch, commit). It ships as a Python-free installer for macOS, Linux and Windows.
Which languages are supported?: The site and console are available in English, Simplified Chinese, Traditional Chinese, Japanese and Korean, switchable at any time.
Can I try it for free?: Yes. Signing up includes a free starter seat, so you can connect a data source and see anomaly detection and diagnosis reports. Upgrade to a paid tier when you need more seats or task quota.

Start free (first month free)