binance-square

binance-square is a skill for Claude Code from aAAaqwq/AGI-Super-Team. It costs 102 tokens per session (6,210 once invoked), scanned A, original, MIT.

A read-only radar for speculative Binance futures trading opportunities. It checks recent Binance Square posts against market data and technical indicators, then creates a local report with its evidence and status.

In plain words
What is it for?
Use it to scan Binance Square posts, verify candidate signals with prices and chart indicators, and review the resulting JSON or Markdown report.
Why use it?
It brings social posts and market measurements into one auditable report while making clear that the tool does not place trades or treat incomplete results as executable.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin.

Part of the agi-super-team plugin — 193 skills, 1 agent shipped together

Good fit Use it to scan Binance Square posts, verify candidate signals with prices and chart indicators, and review the resulting JSON or Markdown report.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/aaaaqwq/agi-super-team/binance-square
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add aAAaqwq/AGI-Super-Team --skill binance-square
Clone the repo
git clone --depth 1 https://github.com/aAAaqwq/AGI-Super-Team

Made for: Claude Code.

Or install agi-super-team, the plugin that ships this one along with the rest of its 193 skills, 1 agent.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for binance-square

README.md
[![agentmods](https://agentmods.dev/badge/skills/aaaaqwq/agi-super-team/binance-square/github.svg)](https://agentmods.dev/skills/aaaaqwq/agi-super-team/binance-square)
Your own site
<a href="https://agentmods.dev/skills/aaaaqwq/agi-super-team/binance-square"><img src="https://agentmods.dev/badge/skills/aaaaqwq/agi-super-team/binance-square/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for binance-square

Your own site · 80×15
<a href="https://agentmods.dev/skills/aaaaqwq/agi-super-team/binance-square"><img src="https://agentmods.dev/badge/skills/aaaaqwq/agi-super-team/binance-square.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 102 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 6,210 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00102 $0.06210
Opus 5 $0.00051 $0.03105
Sonnet 5 $0.00020 $0.01242
Haiku 4.5 $0.00010 $0.00621

Measured 4d ago against content hash fbc2f14168b3, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

binance-square scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.

The scan reads SKILL.md. This mod also ships 37 executable files (scanner/__init__.py, scanner/authors.py, scanner/binance_news.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/binance-square/SKILL.md · 273 lines

How it starts

The opening of the file, as written. The whole thing — 273 lines — stays where its author put it; the contents beside it link to each section on GitHub.

币安广场合约投机雷达 v5.0(Shadow)

当前可真实运行的是只读影子报告:不发 Telegram、不创建定时任务、不下单。

必读合同

执行前完整读取:

  1. GRILL_DECISIONS.md:用户确认的产品、作者、行情、评分、追踪与留存规则。
  2. V4_IMPLEMENTATION_PLAN.md:里程碑、证据门和已知阻塞。

不得把设计目标描述成已经上线的能力。最终结论必须来自当次生成的 JSON/Markdown 和测试结果。

当前已实现

  • v5.0 对候选分层做了探索性放宽:TOP >= 55WATCH >= 40;缺少 Entry/SL/TP 时按字段完整度部分计分,不再仅因缺字段进入评分硬过滤,但执行就绪状态保持 UNAVAILABLE,不能成为可执行 TOP。TP1 RR 1.2 与主要目标 RR 1.5 是评分线;主要目标低于 1.5 仍硬过滤。

  • v5.0 增加 --scheduled-retry 入口和最多两次相互独立的 Feed 捕获;重试不得跨次拼接帖子或复用旧快照冒充本轮成功。

  • Telegram renderer 改为面向用户的简洁摘要;完整 JSON/Markdown、manifest 与 SQLite 仍是唯一审计事实源,精简消息不能替代证据文件。

  • 标准入口默认容量 200;Profile 是专业作者帖主通道,Feed 与经过人工/模型提取的参数信号 URL 作有界补充,按 canonical post_id 去重且保留全部来源 observation。

  • Feed 使用 square-feed-coverage/v1:必须从顶部开始,并记录滚动几何、两相懒加载触发、停止原因和覆盖状态。当前 scope 固定为 BINANCE_SQUARE_DISCOVER_DOMglobal_denominator_known=falsepagination_api_exhaustion_verified=false;即使达到声明下限并稳定穷尽 DOM 而标 BOUNDED_COMPLETE,也只代表该次 DOM surface,不代表内部 feed API 或整个平台。容量、滚动预算、错误或未验证旧快照分别标 CAPPED/PARTIAL/BLOCKED/LEGACY_UNVERIFIED

  • Feed 每次成功观察都写不可变快照和 binance_raw_posts.json observed latest。capture_attempt_nocapture_attempt_limit 与表示稳定物理底部事实的 surface_exhaustedsquare-feed-coverage/v1 的向后兼容可选诊断扩展,旧 v1 证据不会被静默升级。只有严格满足 BOUNDED_COMPLETE + EXHAUSTED + minimum_target_met + unique>=preferred_minimum 的 Discover DOM 捕获才推进独立 binance_raw_posts_eligible.json。PARTIAL 证据必须保存但不得覆盖既有 eligible 指针;抓取锁忙以非零状态失败,不能复用旧 latest 冒充本轮成功。

  • Feed 文本在 CDP 入站时执行 Unicode scalar 规范化:合法 emoji 代理对保持原义,孤立 UTF-16 surrogate 仅替换为 U+FFFD 并计入 malformed_unicode_replacements。最终 JSON 写入层再次递归兜底;写入失败必须清除临时文件,单个异常字符不得拖垮整批帖子。

  • 生产 wrapper 默认最多执行两次相互独立的 Feed 捕获(--capture-attempt-limit 1 可回滚为单次),首次 eligible 立即停止;两次都未 eligible 时只选择最新一次合法捕获进入 radar,不跨次拼接帖子,也不读取旧 eligible。PARTIAL 与达到 200 条硬上限的 CAPPED 都只作带原始覆盖标签的不完整 Feed 补充,CAPPED 不得冒充穷尽。每次 observed latest 都必须刷新、与当次 immutable snapshot 逐字节一致并通过 v1 合同;只有 eligible 捕获才额外要求 eligible pointer 逐字节一致。Feed 是 Profile 主通道的补充,合法不完整捕获不再阻断 Profile,但下游仍必须如实保留覆盖状态;malformed、未刷新、错误或指针不一致继续失败关闭。

  • 逐帖请求币安公开详情接口,使用稳定 post_idauthor_id、权威正文与精确发布时间。

  • Profile + Feed 双通道已接入真实只读 pipeline:Profile 计划只认稳定 squareUid,通过匿名公共 GET 从 timeOffset=-1 分页;逐条用 firstReleaseTime 执行严格24小时窗口,并以可验证 cursor 水位、空页或无 next 作为终止证明。置顶旧帖不会造成提前停止,cursor 停滞/上升、锚点不匹配、schema 漂移、请求失败或页预算耗尽均失败关闭为 PARTIAL

  • 历史 fixture 保留 30D PNL/ROI/VOLUME/WIN_RATE 四榜解析合同;当前真实 Smart Money 接入只支持官方 Web UI 已观测的 PNL/ROI,每榜必须恰好30个唯一排名与 topTraderId 才可标记 COMPLETE

  • Smart Money 公共只读 API 已接入标准 pipeline:--smart-money 在线采集,--smart-money-fixture 注入带完整性校验的证据;排行、交易员详情与 Square 身份映射分别计数,不能相互代替。

  • 单次生产影子入口 scripts/run_production_cycle.py 会按上述独立捕获规则刷新 Feed,再以前台、顺序、--no-send 方式运行 Smart Money、官方新闻与雷达;Feed 快照超过10分钟或来自未来时失败关闭。离线测试只验证选择与失败关闭合同,尚无真实双捕获稳定性回执。

  • 生产 wrapper 为每个长时间子进程每30秒输出一次可刷新的 [HEARTBEAT],避免调度器把正常静默分析误判为无输出挂死;父进程异常退出时会终止并回收子进程。run_radar.pySIGTERM 转成可审计异常,使已创建的 attempt 进入 FAILED,而不是永久遗留 RUNNING

  • scripts/render_latest_telegram.py --report <committed-report.json> 只读取已提交报告并复用标准 render_telegram 合同输出精简文本;它本身不发送。账号、目标、调度与实际发送仍属于安装环境的外部配置,不随公共 Skill 发布。

  • 生产入口默认不再消费旧版可变 data/signal_check_input.json;只有显式传入 --signals-json 才把它作为有manifest的补充来源。真实 pipeline 在任何 Profile、Smart Money 或帖子详情网络请求之前获取一次 Futures catalog 并完成 Binance server-time 校验,后续行情复用同一 catalog。

  • 官方 Binance 公告可用 --official-news 纳入严格前24小时来源覆盖;列表缺发布时间时必须回源详情,响应 envelope、文章 code、新到旧排序和窗口穷尽性均受合同约束,限流或解析失败会留下 FAILED 来源状态而不是伪造空新闻。

  • Smart Money fixture 导入不只校验可重签的 payload hash,还会重新验证固定语义:30DDESC、恰好 PNL+ROI、每页10行×3页,以及两个 onlyShow* filter 均为 false;任一不符即拒绝。

  • topTraderId → squareUid 身份映射 v2 必须来自同一匿名 Binance 响应,并绑定 raw bytes、request manifest、实际 SHA-256 和 /data/topTraderId/data/squareUid JSON Pointer。采集器只能生成 PROPOSED;只有经独立、显式 review 生成的不可变 APPROVED artifact 才能进入活动投影,且支持事件化撤销。

  • 显式映射证据或经哈希固定的多映射 catalog 可通过 --smart-money-square-mapping-evidence 接入同一 pipeline;报告把 Smart Money 映射/Profile 与独立 seed Profile 分开记分母,seed 覆盖不得冒充60人榜单覆盖。

  • 严格最近 24 小时窗口:[decision_at - 24h, decision_at)

  • 旧 signal 文件只提供作者原始 Entry/SL/TP 等参数;时间与作者必须回源重验,标记为 AUTHOR_LEGACY_EXTRACTED

  • DERIVATION_POLICY_V1:作者参数有效时保留 AUTHOR;缺参数或入场已错过时,只有完整 Futures 四周期证据、结构、0.4 ATR 止损、RR 1.5/2.0 与禁追价门全部通过,才生成显著标记的 SYSTEM_REDERIVED;否则 REJECTED

  • 作者原始计划会从发布时间回放到决策时点;期间先触发 SL/TP 即拒绝,同根 K 线同时覆盖 SL/TP 时保守按止损优先。作者止损还必须通过发布前结构和最小 0.4 ATR 距离门。

  • 布林带收缩但尚未放量突破时降为 WATCH;只有方向一致且量能确认的突破才允许继续参与 TOP 判定。

  • 共识按 symbol/direction、稳定作者ID与内容哈希建立;候选自身、同作者多帖和复制内容簇不增加共识分。

  • 未验证作者不冒充 Tier A,作者可信度为 0 分。

  • 精确验证 TRADING / PERPETUAL / USDT 合约。

  • 每次真实运行保存完整 Futures 合约目录证据,不只保存本轮涉及的币种。

  • Futures 公共行情:现价、标记价、指数价、资金费率、持仓量、24h量、15m/1h/4h/1d 已收盘 K 线。

  • Futures 暂时不可用时,只允许 Spot Proxy 补价格和 OHLC;不得用 Spot 量冒充合约量。

  • 指标:BB20、%B、BandWidth、ATR14、Futures volume ratio、多周期趋势。

  • 固定 100 分评分、硬门槛、TOP3/WATCH5、同币多空冲突降级。

  • 手续费、滑点、资金费率和深度模型默认未配置;只有完整注入成本参数且成本调整后主要目标 RR >=2 才可标记执行就绪,不能把名义 RR 报成净 RR。

  • 本地不可变 raw/market/report JSON、Markdown、单一 latest 目录指针原子替换、SQLite logical run/attempt/manifest。

  • 生产 logical run 仍由固定 UTC 四小时 slot 唯一标识;首次 attempt 的 decision_at 取 attempt-start UTC,并必须由 Binance server time 在30秒偏差内验证、且位于 slot 后 0–10 分钟。验证失败则 market_catalog/FAILED,禁止用本机时钟降级成功。SQLite production_run_cutoff 每轮只写一次;唯一一次 +10 分钟 retry 必须精确复用,不能改写 cutoff。帖子资格与已闭合 K 线以它为 cutoff;标量行情是随后抓取的实时验证值,必须单独显示 captured_at

  • job_namespace + production_job_id + scheduled_for_utc 共同隔离 production/canary 身份,避免不同账本生成相同逻辑运行或 attempt 标识。

  • logical run 与 RUNNING attempt 在任何输入文件、catalog 或帖子详情采集之前落库;FETCH/分析/报告失败均终结为带阶段的 FAILED,retry只能从一个已审计的 FAILED attempt1 分配 attempt2。

  • 生产跨轮去重只读取更早且 SUCCEEDED 的同 job 时槽;FAILED/RUNNING 不污染基线。回放默认明确显示 NOT_COMPUTED

  • v4.1 增量数据合同与追踪持久化继续保留:signal family/revision、PENDING/触发/结果、MFE/MAE/R;当前轮失败后查询自动排除其状态。

  • UTC 六时槽、一次 +10 分钟重试、静默恢复与二次失败分类告警已有纯 shadow runner;重试成功仍生成该轮唯一的常规 HELD/NO_SEND 报告 intent,但不生成恢复告警;不安装任务、不发送。

  • 报告分别记录广场抓取开始/完成、行情最新和报告生成时间;K线请求冻结 endTime=decision_time_ms-1,跨周期边界仍保持点时一致。

  • 报告固定对账 Feed、Profile、Smart Money、官方新闻、行情目录五类来源;每类都显示 COMPLETE/PARTIAL/FAILED/NOT_ATTEMPTED、provenance、采集时间和证据清单,不能用一个来源替代另一个来源的完成度。

  • Profile报告必须同时携带Smart Money与seed cohort分母;planned、五类终态、逐作者outcome、稳定身份、顶层状态与source coverage必须对账且唯一。覆盖率分子仅为COMPLETE+EMPTY,任何PARTIAL都不算完成;全EMPTY表示已穷尽无帖,归一为COMPLETE

Read the full file on GitHub · 273 lines

Files

What ships with it

60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 4d ago First seen · 273 lines · 102 tokens per session scan A fbc2f14168b3

Subscribe to this mod's changes

binance-square is a skill published in the GitHub repository aAAaqwq/AGI-Super-Team (91 stars, last pushed 2d ago), licensed MIT. It adds 102 tokens to every session and 6,210 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.