AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Threads Filter

skill-gaojiongwenv587-beep-threads-skills-threads-filter · by gaojiongwenv587-beep

Threads 智能篩選 Skill。三源採集(Feed+關鍵詞+對標帳號)+ 三維熱度評分(互動/跨源/時效)+ AI雙關篩選(排除詞→語境判斷)。觸發詞:篩選帖子、智能篩選、找目標帖子、哪些帖子適合評論、評論目標篩選、抓帖篩選、先篩選再評論、find posts to comment、filter posts、三維評分、熱度評分。輸出按優先級排序的評論候選列表。

No reviews yet
0 installs
0 views
view→install

Install

$ agentstack add skill-gaojiongwenv587-beep-threads-skills-threads-filter

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-gaojiongwenv587-beep-threads-skills-threads-filter)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Threads Filter? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

threads-filter — 智能篩選 Skill

> 三源採集 → 存檔 → filter-comment.py 三維評分 → 輸出評論候選列表

評分邏輯是真正的 Python 代碼,不是提示詞估算。 篩選腳本位於:~/Desktop/threads-filter-comment/filter-comment.py


PHASE SETUP:首次配置向導

觸發條件~/.threads-filter-comment.json 不存在,或用戶說「重新配置篩選」。

向導流程

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🎯 threads-filter 首次配置向導
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Step 1/4:關鍵詞矩陣(keywords / priority_keywords)
Step 2/4:排除詞庫(exclude_keywords)
Step 3/4:AI 配置(api_url / api_key / model)
Step 4/4:確認並寫入
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Step 1 / 4 — 關鍵詞矩陣

請提供兩類關鍵詞:

🔑 高優先核心詞(命中即標記 high priority):
   例:韓國、首爾、江南、釜山、韓國醫美、飛韓國

🏷️ 一般關鍵詞(命中標記 medium priority):
   例:醫美、整形、微整、玻尿酸、保養、護膚、外貌焦慮

→ 儲存為 priority_keywordskeywords


Step 2 / 4 — 排除詞庫

哪些帖子要直接跳過?
預設已包含:醫院、診所、歡迎預約、歡迎諮詢、價格、優惠、促銷、line:、微信
需要補充同業競品名稱嗎?

→ 儲存為 exclude_keywords


Step 3 / 4 — AI 配置

是否啟用 AI 語境判斷?(預設:啟用)
如果啟用,請提供:
  ai_api_url:(OpenAI 相容端點)
  ai_api_key:
  ai_model:(預設 Qwen/Qwen3.5-27B-FP8)

Step 4 / 4 — 寫入配置

配置寫入 ~/.threads-filter-comment.json

{
  "ai_enabled": true,
  "ai_api_url": "https://...",
  "ai_api_key": "sk-...",
  "ai_model": "Qwen/Qwen3.5-27B-FP8",
  "keywords": ["醫美", "整形", "保養", "護膚", "外貌焦慮"],
  "exclude_keywords": ["診所", "歡迎預約", "促銷", "line:"],
  "priority_keywords": ["韓國", "首爾", "江南", "釜山", "韓國醫美"]
}

使用方式

執行篩選(三源)     → 「幫我篩選適合評論的帖子」
只用 Feed 篩選       → 「只抓首頁 Feed 篩選」
關閉 AI 快速篩選     → 「不用 AI,只做關鍵詞篩選」
指定帳號             → 「用 account2 篩選帖子」
重新配置             → 「重新配置篩選」

PHASE 1:三源採集 → 存至暫存文件

三個來源分別採集,存為獨立 JSON 文件,供 filter-comment.py 讀取。

ACCOUNT="default"   # 或用戶指定帳號
TMPDIR="/tmp/threads-filter"
mkdir -p "$TMPDIR"

來源 A:首頁 Feed(50條)→ /tmp/threads-filter/feed.json

uv run python scripts/cli.py --account "$ACCOUNT" list-feeds --limit 50 \
  > /tmp/threads-filter/feed.json

來源 B:關鍵詞搜索 → /tmp/threads-filter/keyword.json

讀取配置中 priority_keywords + keywords,逐一搜索(每個取最新 20 條),合併輸出:

# 對每個關鍵詞執行:
uv run python scripts/cli.py --account "$ACCOUNT" \
  search --query "[關鍵詞]" --type recent --limit 20
# 將所有結果的 posts 陣列合併,寫入 /tmp/threads-filter/keyword.json

若關鍵詞較多(>5個),逐一執行後在內存合併,最終輸出格式:

{"posts": [ ...所有帖子... ]}

來源 C:對標帳號(可選)→ /tmp/threads-filter/benchmark.json

若用戶提供對標帳號列表,抓取每個帳號近期 15 條:

uv run python scripts/cli.py --account "$ACCOUNT" \
  user-profile --username "@帳號名" --limit 15
# 合併後寫入 /tmp/threads-filter/benchmark.json

PHASE 2:呼叫 filter-comment.py 執行三維評分

採集完成後,呼叫真正的 Python 評分腳本:

FILTER_SCRIPT=~/Desktop/threads-filter-comment/filter-comment.py

# 三源模式
python3 "$FILTER_SCRIPT" \
  --feed-file      /tmp/threads-filter/feed.json \
  --keyword-file   /tmp/threads-filter/keyword.json \
  --benchmark-file /tmp/threads-filter/benchmark.json \
  > /tmp/threads-filter/result.json

# 僅 Feed 模式
python3 "$FILTER_SCRIPT" \
  --feed-file /tmp/threads-filter/feed.json \
  > /tmp/threads-filter/result.json

# 關閉 AI(快速模式)
python3 "$FILTER_SCRIPT" --no-ai \
  --feed-file      /tmp/threads-filter/feed.json \
  --keyword-file   /tmp/threads-filter/keyword.json \
  > /tmp/threads-filter/result.json

三維評分說明(代碼實現,非估算)

| 維度 | 滿分 | 計算方式 | |------|------|---------| | 互動分 | 40 | 歸一化(點贊 + 回覆×2 + 轉發×3),以本批次最高值為基準 | | 跨源分 | 35 | 按來源組合給分(見下表) | | 時效分 | 25 | 按發帖時間衰減(見下表) | | 綜合分 | 100 | 互動×0.4 + 跨源×0.35 + 時效×0.25 |

跨源分對照:

| 出現來源 | 得分 | |---------|------| | 三源都出現 | 35 | | Feed + 對標帳號 | 22 | | Feed + 關鍵詞 | 20 | | 僅 Feed | 10 | | 僅關鍵詞 | 8 | | 僅對標帳號 | 8 |

時效分對照:

| 發帖時間 | 得分 | |---------|------| | 0–6 小時 | 25.0(滿分) | | 6–24 小時 | 16.7 | | 24–48 小時 | 10.0 | | 48 小時以上 | 3.3 |


PHASE 3:解析結果並呈現

讀取 /tmp/threads-filter/result.json,按以下格式呈現給用戶:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🎯 篩選完成(YYYY-MM-DD HH:mm)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
採集來源
  Feed          XX 條
  關鍵詞搜索    XX 條
  對標帳號      XX 條
  去重後總計    XX 條

關鍵詞篩選後   XX 條候選
AI 語境判斷後  XX 條適合評論

優先級分布
  🔴 高優先(priority: high)  X 條
  🟡 中優先(priority: medium)X 條

建議本次執行:取高優先全部 + 中優先前 N 條
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

高優先帖子:
1. [@用戶] 帖子摘要(30字內)
   綜合分:XX|互動:XX|跨源:XX|時效:XX
   來源:feed + keyword
   URL: https://...
   AI評論:「...」

2. ...

結果 JSON 欄位說明:

{
  "total_input": 120,
  "total_deduplicated": 95,
  "total_filtered": 8,
  "results": [
    {
      "post": { "postId": "...", "content": "...", ... },
      "sources": ["feed", "keyword"],
      "priority": "high",
      "match_reason": "韓國/首爾相關",
      "score_total": 72.3,
      "score_interaction": 35.0,
      "score_cross_source": 20.0,
      "score_timeliness": 25.0,
      "ai_should_comment": true,
      "ai_comment": "...",
      "ai_reason": "..."
    }
  ]
}

與其他 Skill 的配合

篩選完成後 → 交給 threads-interact 執行 reply-thread 評論
篩選結果   → /tmp/threads-filter/result.json 可直接作為下一步輸入
配置管理   → 說「重新配置篩選」進入向導

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.