Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计提醒

x-opsx 操作

Agent Skill

x-ops 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

742

周安装

30

GitHub Stars

公开资料未说明

下载量

233
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:x-ops(x 操作)
来源仓库:https://github.com/saltbo/agent-skills
仓库路径:skills/x-ops
安装命令:
npx skills add https://github.com/saltbo/agent-skills --skill x-ops
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/saltbo/agent-skills --skill x-ops

简介

x-ops 用于查找、检索和筛选相关信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词或任务场景快速定位候选结果。
  • 通过 npx skills add 命令从指定 GitHub 仓库安装使用。
  • 需确认权限范围和维护状态,注意是否触发联网、命令执行或文件读写操作。
  • x-ops 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

X Ops — Operational Procedures

All operations use x-cli.py (tweepy + official X API v2). Set the alias at the start of every session:

X="uv run .agents/skills/x-ops/x-cli.py"

Commands

Read

$X profile <username>          # full profile with followers/following/tweet count
$X tweet <tweet_id>            # tweet with reply_count, likes, views, author_followers
$X search "<query>" [count]    # search recent tweets (min 10, max 100)
$X mentions <user_id> [count]  # mentions of a user

All read commands return JSON with full metrics (reply_count, followers, etc).

Write

$X post "<text>"               # post a tweet
$X reply <tweet_id> "<text>"   # reply to a tweet
$X like <tweet_id>             # like a tweet
$X follow <user_id>            # follow a user
$X unfollow <user_id>          # unfollow a user

Write commands return JSON confirmation with the created tweet ID.

Rate Limits (X API v2)

OperationLimit
search60 / 15min
mentions10 / 15min
profile95 / 15min
tweet300 / 15min
post100 / day
like1000 / day
follow15 / 15min

Wait 2-3 seconds between write operations.

API Constraint

X API blocks replies/quotes to accounts that haven't mentioned @rdsaltbo. Do NOT attempt replies to strangers — always 403. Only reply to mentions.

Cycle Workflow

Minimize Bash calls. Every $X call is a tool round-trip. Target: 8-12 calls per cycle.

  1. $X profile rdsaltbo — get follower count, determine tier (1 call)
  2. $X search "<query>" 20 — run 1-2 searches from task description (1-2 calls)
  3. $X mentions 726438383401074688 — check mentions (1 call)
  4. Reply to actionable mentions only — do NOT reply to strangers (0-2 calls)
  5. Like 5-8 posts from search results — pick substantive posts from target developers (5-8 calls, can batch quickly)
  6. Follow 3-5 developers from search results — use author_id from search data directly (3-5 calls)
  7. Post 1 original tweet based on what you observed in search (1 call)
  8. Log actions + create next task (2 calls)

Search Query Rotation

Use 1-2 per cycle. These find both content inspiration and like/follow targets:

  • claude code
  • AI coding agent
  • cursor AI OR codex
  • multi-agent AND code
  • vibe coding
  • AI developer tools
  • coding agent workflow
  • "build in public" AI tools
  • "looking to connect" developer AI
  • "indie hacker" AI agent

De-duplication

Do NOT like/follow the same person more than once per cycle. Check the task description's "Follow-up from last cycle" — if you already followed someone in previous cycles, skip them.

Task Standards

Title format

x-ops #<seq>: <focus keyword>

Description template

## Search queries
- <query 1>
- <query 2>

## Tweet plan
- Type: <hot take | build-in-public | tip | comparison | observation>
- Topic: <what to write about>
(or "None — replies + connect only cycle")

## Follow-up from last cycle
- <conversation to continue, or "none">

## Notes from last cycle
- <observations about what worked or didn't>

Self-continuation

⚠️ CRITICAL: You MUST include --scheduled-at. Without it, the task runs immediately and creates an infinite loop that spams the account. This has happened before — do NOT skip this parameter.

Calculate the scheduled time with a random offset (avoid exact intervals — looks robotic):

OFFSET=$((110 + RANDOM % 40))m  # 110-150 minutes (roughly 2h with jitter)
NEXT=$(date -u -v+${OFFSET} +"%Y-%m-%dT%H:%M:%SZ")

# Check if NEXT falls in quiet hours (UTC 06:00-14:00 = EST 01:00-09:00 = PST 22:00-06:00)
# If so, push to 14:00 UTC (09:00 EST / 06:00 PST)
NEXT_HOUR=$(date -u -v+${OFFSET} +"%H")
if [ "$NEXT_HOUR" -ge 6 ] && [ "$NEXT_HOUR" -lt 14 ]; then
  NEXT=$(date -u -v+1d +"%Y-%m-%dT14:%M:%SZ")  # next day 14:00 UTC
fi
ak create task \
  --board jb21kfv6 \
  --assign-to e6f896b845f81e93 \
  --scheduled-at "$NEXT" \
  --title "x-ops #<next-seq>: <focus>" \
  --description "<filled template>" \
  --priority medium \
  --labels "ops,engagement,connect"

Verify the created task has a scheduled_at value in the response. If not, cancel it immediately.

Comment Standards

Post comments using ak task log <task-id> "<message>":

1 — Start

CYCLE START | Followers: <n> | Tier: <phase> | Queries: <q1>, <q2>

2 — Likes

LIKES (<count>)
♥ @<user> (<followers>) — "<first 50 chars of their post>" [tweet:<id>]

3 — Follows

FOLLOWS (<count>)
+ @<user> (<followers>) — <reason>

4 — Mentions

MENTIONS (<count> new)
↩ @<user> — "<reply preview>" [tweet:<id>]
— no actionable mentions

5 — Tweet

TWEET [<id>]
"<full tweet text>"
Type: <type>

6 — Summary

CYCLE COMPLETE | Followers: <n>
Likes: <n> | Follows: <n> | Mentions replied: <n> | Tweet: <0 or 1>
Next: <task-id> scheduled <time>

Error Handling

  • 403 on reply → skip target, find another. Do NOT count as success.
  • Rate limit error → wait until reset, then retry.
  • Auth error → stop and report. Do not retry.
  • If ak create task fails → retry once after 10 seconds.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.41%
按下载量换算85

Claude

29.41%
按下载量换算69

Cursor

19.36%
按下载量换算45

Gemini CLI

9.72%
按下载量换算23

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills