Ai.300Live

This week imported

0

Outreach Funnel

0

First Outreach

›

0

Opened

›

0

Replied

Start Collaboration

0

Total Subscribers

0

All creators in DB

Country Source Top 5

By imported creator count

Subscriber Segment Distribution

Creator Count × Subscriber Tier

0

110K-1010K

0

1010K-5010K

0

5010K-10010K

0

10010K-50010K

0

50010K+

Engagement Rate × View rate bubble chart

Bubble Size = Subscribers；Color-code content types

Videos

Shorts

Creator Tier Assignment

by Content Type / Category，Determine placement priority

Source Type

Mostly long videos

Main Pool

ShortsShare < 30%

0 creators

Share in DB 0%

Source Type

Shorts-oriented

Shorts pool

ShortsShare > 70%

0 creators

Share in DB 0%

Source Type

Mixed / Low Activity

Low health

Published (30d) < 2

0 creators

Share in DB 0%

Selected 0

	Channel	Country	Import Source	Content Direction	Creator Type	Subscribers	Engagement Rate	View Rate	Import Date	Email	Emails Sent

0 of 0

YouTube Creator Screening

Videos / Shorts dual mode · Local storage · Real-time params

API Settings

YouTube Data API v3 Key

💡 Daily free quota 10,000 units, resets at 00:00 Pacific Time.12keywords × 4 pages ~ 5,800 Unit。

API Call Flow & Quota Usage Details

Step	API Endpoint	Quota/Call	Channel A	Channel B	Description
1.Search & Collect	videos.list (chart=mostPopular)	1/page	4~8	—	Channel A: Fetch trending game videos by region
1.Search & Collect	search.list (type=video)	100/page	—	1200~4800	Channel B: Search videos by keywords
2.Coarse Screen	channels.list	1/50Channel	≈3~4	≈6~12	Batch fetch subs/country, eliminate 50-70%
3.Deep Analysis	playlistItems.list	1/Channel	≈1/Channel	≈1/Channel	Only channels passing coarse screening
3.Deep Analysis	videos.list	1/Channel	≈1/Channel	≈1/Channel	Fetch plays/Likes/Comments/Brand Collab
Channel A Total（TrendingScan）			≈300~500	—	4page trending≈8 + Channel processing≈300~400
Channel B Total（Keyword Search）			—	≈3,000~6,000	12 keywords × 4 pages≈4800 + Channel processing

💡 Coarse screening flow：Subscribers → Registered Country（Null pass）→ Game Type(topicDetails，Null pass）→ Only passed channels enter deep analysis

Filter Params

Subs Lower Limit

Subs Upper Limit

Avg Views≥

Engagement Rate%

View Rate%

Search Pages

Registered Country

ChannelLanguage

· Channel language affects search relevance
· Registered country is filled at registration
· This field is filled by the creator，May not match actual location

💡 Search channel description

🔥 Channel A（Trending video scan）：Fetch trending game videos from target region channels directly，regionCode Region filter effective，nokeywords required，Very low quota usage（~ 408 Unit），Suitable for quick discovery of trending creators。Cover~ 150-200 channels。
🔍 Channel B（Precise keyword search）：Search videos by keywords，Extract channel info from videos，High accuracy but high quota consumption（~ 5,224 Unit），Suitable for deep vertical discovery。Search Phase regionCode Keyword Searchno。
Suggest：Run Channel A first for quick coverage，Then Channel B for precise deep discovery，Results auto-deduplicated & merged。Only one channel can run at a time。

📊 Data Analysis Base

The base for all data analysis is maxResults = 50 Videos，Filtered out Shorts（≤60sec）

Search keywords（1 per line, changes take effect on next search）

Last search records

No search records

Brand collab detection keywords（Match title/Describe ad markers，reference only）

0

Scanned

0

Passed

0

Does not match

0

Has collaboration records

0

Quota Usage

Real-time Log

Passed (0) Does not match (0) All (0)

	#	Channel Name	Subscribers	Avg Views	View Rate	Engagement Rate	Latest Published	Country	Email	keywords	Brand	Status
No data

	#	Channel Name	Subscribers	Avg Views	View Rate	Engagement Rate	Completion Rate	Frequency	Latest Published	Country	Email	keywords	Brand	Status
No data

Conversion ParamsSettings （Auto-recalculate after modification）

Click Conversion Rate Default 1%

Click User = Avg Views × Click Rate

Download Conversion Default 10%

Downloads = Click User × Download Rate

Estimated Unit Price (USD) Default 5

Total effect = Downloads × Unit Price

Risk Range Default 85%

Suggested Amount = Total effect × Risk Range

Click User = Avg Views × Click Rate → Downloads = Click User × Download Rate → Total effect = Downloads × Unit Price → Suggested collab amount = Total effect × Risk Range

Creators with email：0 name · Checked：0 name

Also send to test mailbox：[email protected] When checked, emails also go to test mailbox

	#	Channel Name	Subscribers	Avg Views (10)	Engagement Rate	Email	Source	Status	Suggested collab amount (USD)
No creator data，Please select creators in Creator DB first, then click「Initiate Outreach」

📧 Email workspace

📮 Tencent Enterprise Mail SMTP Disconnected

▾

Sender email account

SMTP Auth Code

🏢 Server：smtp.exmail.qq.com · Port：465（SSL）
In Tencent Enterprise Mail「Settings → Client-sideSettings」Enable SMTP，Login with auth code

🔍 Self-hosted email tracking（Optional） Not enabled

▾

Enable self-hosted email tracking

Emails still via Tencent Enterprise Mail SMTP Send。
After enabling，Self-hosted tracking service will inject tracking ID into each email，Record open & click behavior in real-time。

📮 Tencent SMTP Send 🔍 Self-hosted Tracking Layer

Closed

✏️ Email template editor

▾

Email Subject

Email Body · [Creator Name] Placeholders auto-replaced at send time

Hi [Creator Name],

I've been watching your game videos, and I really love your content style and energy.

We're looking for creators like you to make a paid gameplay / review video for our mobile game—Starparks.

What we offer:

Full creative control (keep your own style)
Free game support
No strict script

If you're interested, I can send you the full collaboration details right away.

Best regards,
Kxun
CEO
300 LIVE

📤 Email Send

▾

📮

SMTP Direct (Tencent Enterprise Mail)

Emails will be via Tencent Enterprise Mail SMTP directlySend，Without tracking

0

Pending

0

success

0

failed

Ready, check target creators before sending

📊 Self-hosted Tracking Delivery Stats Analytics

▾

📊

Self-hosted tracking not enabled

Enable "Self-hosted Email Tracking" above and complete setup to see data here

📅 Send Records（By Date）

▾

No send records

Account

API Config

Data Refresh

Email Tool

Notifications

Member Management

API Config

YouTube Data API Key

Used for channel data fetching，Daily quota 10,000 units

Daily refresh ratio

Daily refresh ratio of existing creators in the DB

Refresh Time（UTC）

Daily auto-refresh execution time

QuotaMonitor

Quota alert threshold

Notify when usage exceeds this ratio

Disable search.list

Force disable high-consumption endpoints（100u/times）

Email tool sync

Webhook URL

URL for email tool to push status changes

AutoSync

Open/Reply status auto-updates to Creator DB

Timezone

All timestamps based on this timezone

System Plans · Storage Plan

Storage centralization plan — Migrate all storage from scattered local to Tencent Cloud PostgreSQL

Planned

1、Current storage overview

Tencent Cloud Server（mailer300 / 43.130.49.17）

Service	Port	Storage Location	Type	Data Volume
ai.300live.com Frontend	443	/var/www/ai-tools/ + localStorage	Static HTML + Browser Storage	1.4M
storage-api（Creator DB Proxy）	3082	/var/www/storage-api/outreach.db	SQLite	20K
email-server（Email Outreach）	3080	/var/www/youtube-tools/email-server/data.db	SQLite	4K
track-server（Email Tracking）	3081	/opt/track-server/track.db	SQLite	4K
youtube-storage	3082	/var/www/youtube-tools/data/*.json	JSONFile（30day cleanup）	4.8M
langdetect Model	—	/home/ubuntu/langdetect/lid.176.bin	2baseModel	125M

Mac Mini（Tailscale: 100.68.146.94）

gamecreator-hub :3000 — PostgreSQL Creator DB Core Data，Current 13 CreatorRecord
⚠ Mac Mini Is the sole data source for Creator DB，Tencent Cloud storage-api Pass Tailscale Intranet proxy access

MacBook Local（Browser localStorage，29Read/write at）

gchub_creators

Creator DB Local Cache

gchub_outreach

Outreach status tracking

email_send_history

Email send records

email_api_token

Email service auth

gchub_channel_videos

Channel video cache

YouTube API Quota

Quota Tracking

Server Resource Overview

mailer300 · 43.130.49.17 · Intel Xeon E5-26xx v4 · 2 vCPU

2026-04-23 Snapshot

Disk 50 GB

Used 9.8G · canuse 38G

19.6%

Log 732M langdetect 126M email-tracking 62M storage-api 29M youtube-tools 45M track-server 19M ai-tools 3M Other ~8.8G

Memory 1.9 GiB

Used 504Mi · canuse 1.2Gi · Cache 1.3Gi

26%

gamecreator-hub

86M

track-server

72M

email-server

72M

storage-api

72M

PM2 daemon (root)

73M

PM2 daemon (ubuntu)

69M

PostgreSQL live300db · 13 MB

16 tables

user_settings

3.2M

filter_results

240K

sent_emails

112K

creators

96K

open_events

80K

Other 11 tables

336K

2、Centralized migration plan（5 phases）

Week 1 Phase 1：PostgreSQL Deploy + Mac Mini Migrate（Highest Priority）

1.1 Tencent Cloud setup PostgreSQL，Create 300live_db Database and user

1.2 Mac Mini Export gamecreator-hub PostgreSQL Data → pg_dump → scp Transfer to Tencent Cloud

1.3 Tencent Cloud execution pg_restore Import creator DB data

1.4 gamecreator-hub Deploy to Tencent Cloud，Modify .env Point to local PostgreSQL

Week 2 First half Phase 2：SQLite → PostgreSQL Migrate

in 300live_db Create unified table structure：sent_emails / webhook_events / open_events / click_events / tracked_links / outreach_replies / youtube_filter_results

Write Node.js Migration script migrates one by one 4 SQLite DB + JSON FileData

Backup all before migration SQLite File（rename .bak），Post-migration row count check

Week 2 Second half Phase 3：Backend service refactoring

storage-api：Remove Mac Mini Proxy，Direct local connection PostgreSQL

email-server：Remove better-sqlite3，Switch to pg drive

track-server：Remove better-sqlite3，Switch to pg drive

youtube-storage：JSON File read/write changed to youtube_filter_results table（30-day retention cleanup logic）

Week 3 First half Phase 4：Frontend localStorage To Cloud

gchub_creators

High Priority → Real-time API query

gchub_outreach

High Priority → outreach APItable

email_send_history

Medium Priority → sent_emailstable

email_api_token

Medium Priority → Server-side session

gchub_tag_params

Low Priority → Keep local

gchub_filter_history

Low Priority → Keep local

Week 3 Second half Phase 5：Nginx Config Adjustment + cleanup

api.300live.com proxy_pass from Tailscale 100.68.146.94 Change pointer 127.0.0.1:3000

Merge duplicate services：track-server / email-tracking / youtube-storage-server / storage-api eachKeep1

cleanup Mac Mini Tailscale Dependency，Can be taken offline after confirming no dependencies

3、Post-migration target architecture

Tencent Cloud mailer300（43.130.49.17）

├── Nginx :443

│ ├── ai.300live.com → /var/www/ai-tools/

│ ├── api.300live.com → 127.0.0.1:3000 （Local，No longer via Tailscale）

│ ├── youtube.300live.com → :3080 / :3082

│ └── track.300live.com → :3081

├── gamecreator-hub :3000（LocalDeploy）

└── PostgreSQL :5432 — 300live_db

├── creators（Creator DB）

├── sent_emails / webhook_events

├── open_events / click_events / tracked_links

├── outreach_replies

└── youtube_filter_results

Mac Mini → Can take offline（No longer depended on by any services）

4、Risk Assessment & Rollback Plan

Risk	impact	Mitigation
PostgreSQL MigrateDatalost	high	Backup all before migration SQLite File，Post-migration row count check
Mac Mini Going offline causes service interruption	high	Run in parallel on Tencent Cloud first，Switch after verification Nginx
localStorage Data not synced	in	Keep localStorage As fallback plan，Progressive Migration
50G Insufficient disk space	in	Currently used 9.3G，PostgreSQL + Data Estimate < 2G，Low Risk
Concurrent write conflict	low	SQLite Originally single write，PostgreSQL Better concurrency support

Rollback Plan：Original files and configs are retained at each stage — SQLite DB File rename .bak，Nginx Modify after config backup，Mac Mini Service does not stop until fully validated。

System Plans · Multi-language verification

Comment Language Verify — Server-side detection scheme（v2）

Planned

1、Architecture Overview

BrowserFrontend Tencent Cloud Server External

─────────── ──────────────── ─────

┌─ gamecreator-hub (:3000)

click"Comment Lang Verify" ──→ │ POST /api/lang-check │

│ 1. call YouTube API Fetch comments │──→ YouTube Data API v3

│ 2. call Python fasttext Detect Language │

│ 3. 60%Rule-based determination → Write PG + Back │

└────────────┬───────────────────┘

│

PostgreSQL

comment_lang / lang_verified_at

Core Changes：YouTube API Call + Language detection is fully server-side，Frontend only sends one request。API Quota：12 units / times

2、Language determination rules（Core）

Condition	comment_lang value	FrontendShow
certainLanguageShare > 60%	thisLanguage Code（like th、en）	Display on right of country code `✅ th`
No Language over 60%	❓	Display on right of country code `✅ ❓`
No comments / CommentsClose	null	Display on right of country code `—`
Not Verified	null（lang_verified_at also empty）	No marker on right of country code

TH Channel，Comments 70%Thai → TH ✅ th

US Channel，45%English+40%Spanish → US ✅ ❓

JP Channel，Not Verified → JP

KR Channel，Comments disabled → KR —

3、DB changes

-- creators Table adds two columns

ALTER TABLE creators ADD COLUMN comment_lang text;

ALTER TABLE creators ADD COLUMN lang_verified_at timestamptz;

col	Type	Description
comment_lang	text	null=Not verified or no comments；Language Code=Over 60%；❓=No language over 60%
lang_verified_at	timestamptz	Last verification time，null=Not Verified

4、Backend addition API（gamecreator-hub :3000）

4.1 Download fasttext Model

cd /var/www/gamecreator-hub

wget -q https://dl.fbaipublicfiles.com/fasttext/supervised-models/lid.176.bin \

-O /var/www/gamecreator-hub/lid.176.bin

4.2 Python Language detection script（/var/www/gamecreator-hub/lang_detect.py）

#!/usr/bin/env python3

import sys, fasttext

model = fasttext.load_model('/var/www/gamecreator-hub/lid.176.bin')

for line in sys.stdin:

t = line.replace(' ',' ').strip()

if t:

pred = model.predict(t, k=1)

print(pred[0][0].replace('__label__',''))

else: print('und')

4.3 POST /api/lang-check — Process Flow

Request Body：{ "channelId": "UCxxxxxx", "apiKey": "AIzaSy..." }

① channels.list → uploadsPlaylistId → ② playlistItems → Latest 10 video IDs → ③ commentThreads → each3（total30） → ④ fasttextDetect → ⑤ 60%Rule-based determination → Write PG

4.4 Backformat

{ "channelId": "UCxxxxxx",

"comment_lang": "th",

"lang_verified_at": "2026-04-23T12:00:00Z",

"detail": { "totalComments":30, "langDistribution":{"th":22,"en":5,"ko":3},

"topLang":"th", "topLangRatio":0.73, "rule":"th 73% > 60% → th" } }

4.5 Exceptions

• CommentsClose / 0 comments → comment_lang = null，lang_verified_at = now()

• YouTube API Error → HTTP 500 + Error Message

• PUT /api/creators allowed Extension：Add comment_lang, lang_verified_at

5、Frontend changes

5.1 New Button

in「Remove」Added after button 「Comment Lang Verify」 Button：

• Default dimmed（Gray semi-transparent），SelectedCreatorthen activate

• Each timeonlyAllowSelected 1 CreatorperformVerify

• Showing during verification loading Rotate icon

5.2「Country」Column display changes（Append marker to right of country code）

Not Verified：TH

Verified with primary language：TH ✅ th

Verified no primary language：TH ✅ ❓

Verified no comments：TH —

5.3 New JS Function langCheckCreator()

① Check selected creator count（Must be exactly 1）

② Fetch selected creator's channelId

③ Select API Key with lowest quota

④ POST /api/lang-check { channelId, apiKey }

⑤ Update local data after success + Re-render list

⑥ Button restore state

6、Execution Steps（12 step）

#	Action	File / Position
1	Download fasttext lid.176.bin Model	Server
2	Create lang_detect.py Script	/var/www/gamecreator-hub/
3	Test Python Script	Server CLI
4	PG Add column comment_lang + lang_verified_at	Server psql
5	server.js Add child_process + detectLanguages()	Server
6	server.js New POST /api/lang-check Route	Server
7	server.js PUT allowed Extension	Server
8	pm2 restart gamecreator-hub	Server
9	Frontend UI + JS Change	ai-tools/index.html
10	SyncChange	youtube-tools/index.html
11	node --check Syntax Validation	Server
12	End-to-end test	Browser

7、Risks & Alternatives

Risk	Response
lid.176.bin Downloadfailed	scp Transfer
YouTube VideoNo comments	comment_lang=null，Show TH —
Comments disabled	same"No comments"process
API Key Quota Exhausted	Frontend selects lowest quota Key，Notify user when exhausted

System Plans · ApiPlan

4core tables × Frontend Feature × APIEndpoint Association Mind Map — Click "Read Field" to view all referencing features

Database 4 core tables verified

channels（Creator Basic Data） 15 Field

channel_id PK · channel_name · country · subscribers · avg_views_30 · play_rate · engagement_rate · last_published_at · has_sponsorship · email · game_type enum · creator_type enum · content_type · activity_30d · imported_at

outreach（Outreach Log） 7 Field

id PK · channel_id →channels · sent_at · opened_at · replied_at · deal_started_at · campaign_id

users（Account System） 5 Field

id PK · email · password_hash · role admin/viewer · created_at

sync_logs（Sync Log） 5 Field

id PK · source email_tool/google_api · synced_at · records_count · status

Database、Frontend feature and YouTube API Endpoint association mind map

clickRead FieldTags → View all frontend features referencing this field | clickFeature row → View data lineage description

channels

outreach

users

sync_logs

Calculate locally（0 units）

System Plans · Creator tag params

Matching keywords for content direction and creator type/Determination Rules — Auto-classify at import，Supports manual correction

Content Direction Game Type — Matching keyword examples

Scan all videos at import（~50），Count keyword-matched videos per category，Tag with the most-matched category。

Creator Type Matching keyword examples

Scan all videos at import（~50），Count keyword-matched videos per creator type，Tag with the most-matched type。

Tagging logic description · Scoring system
At import, for each creator ~ 50 videos scanned one by one，Count per category/Number of keyword-matched videos for this type，Tag with the type having the most matched videos（Rather than tagging on first match）。On tie, take earlier list item。Fallback when no match"Other"。Manual correction supported after auto-tagging。