← 返回
AI智能 中文

Coin Identifier

Identify coins from photos using evidence-based visual checks, ranked candidates, mint-mark reasoning, and a reusable local catalog.
利用基于证据的视觉检查、候选排序、铸造标记推理及可复用本地目录,从照片识别硬币。
ivangdavila
AI智能 clawhub v1.0.0 1 版本 100000 Key: 无需
★ 0
Stars
📥 495
下载
💾 10
安装
1
版本
#latest

概述

When to Use

Use when the user wants to identify a coin from one or more photos, narrow down similar issues, log a collection piece, or separate likely type from later grading or pricing work.

Architecture

Memory lives in ~/coin-identifier/. If ~/coin-identifier/ does not exist, run setup.md. See memory-template.md for structure.

~/coin-identifier/
├── memory.md
├── identifications/
│   └── YYYY-MM/
│       └── {entry-id}.md
└── exports/

Quick Reference

TopicFile
-------------
Setup guidesetup.md
Memory templatememory-template.md
Coin evidence checklistevidence-guide.md

Scope

This skill ONLY:

  • identifies coins from visible evidence in user-supplied images
  • returns ranked candidates with explicit confidence and missing evidence
  • asks for the next best photo or measurement when the evidence is incomplete
  • stores local identification notes only if the user approves

This skill NEVER:

  • guarantee authenticity, grade, mint error status, metal purity, or market value from photos alone
  • recommend cleaning, polishing, or altering a coin
  • upload images or coin data to external services

Security & Privacy

Data stored locally if approved by the user:

  • activation and response preferences in ~/coin-identifier/memory.md
  • one note per saved identification in ~/coin-identifier/identifications/

This skill does NOT:

  • make network requests
  • claim professional grading or authentication
  • write local files without user approval

Core Rules

1. Clear the photo gate before naming a coin

  • Check subject isolation, glare, blur, crop, orientation, and whether the obverse, reverse, or edge are missing.
  • If the coin is angled, reflective, inside a sleeve, or mixed with other coins, ask for a tighter straight-on view first.

2. Return ranked candidates with confidence, not one blind guess

  • Give one to three candidates with confidence bands: High 85-95, Medium 60-84, Low 35-59.
  • For each candidate, cite the visible evidence and the missing evidence.
  • If the signal is weak, say the result is an unresolved shortlist instead of pretending certainty.

3. Use coin evidence in a fixed order

  • Open evidence-guide.md before deciding.
  • Work from country or script, portrait or emblem, denomination, date, mint mark, metal color, shape, rim or edge, then commemorative cues.
  • Keep obverse, reverse, and edge evidence separate.

4. Ask for the next best view, not generic more photos

  • Prefer straight obverse, straight reverse, edge, mint-mark crop, and scale or weight.
  • Explain which missing feature would separate candidate A from candidate B.

5. Separate identification from value, grading, and authenticity

  • Photo identification can narrow the type and likely issue without proving grade, rarity, or authenticity.
  • If the user wants value or authenticity, treat identification as step one and keep the rest provisional.

6. Keep memory useful and lightweight

  • Save only durable preferences and approved identification notes.
  • One saved entry should record date, coin label, best match, confidence, evidence, and unresolved questions.
  • Do not write files unless the user approves local storage.

7. Say what could change the answer

  • Highlight wear, glare, missing edge data, foreign-script ambiguity, and similar commemoratives when they limit certainty.
  • Update the shortlist immediately if a better image or measurement changes the balance.

Common Traps

  • Guessing from one reflective angled photo -> dates, mint marks, and legends disappear.
  • Treating any silver-colored coin as silver bullion -> composition and coin type get conflated.
  • Calling a commemorative theme the country or denomination -> wrong catalog family.
  • Jumping from identification to market value -> grade, authenticity, and demand remain unverified.

Related Skills

Install with clawhub install if user confirms:

  • image - inspect and optimize photos before identification
  • image-edit - crop, isolate, and clean up the subject for clearer review
  • inventory - maintain a broader catalog once coins are identified
  • scanner - improve flat top-down captures of coins, cards, or documents

Feedback

  • If useful: clawhub star coin-identifier
  • Stay updated: clawhub sync

版本历史

共 1 个版本

  • v1.0.0 当前
    2026-03-30 04:32 安全 安全

安全检测

腾讯云安全 (Keen)

安全,无风险
查看报告

腾讯云安全 (Sanbu)

安全,无风险
查看报告

🔗 相关推荐

ai-intelligence

Proactive Agent

halthelobster
将AI智能体从任务执行者升级为主动预判需求、持续优化的智能伙伴。集成WAL协议、工作缓冲区、自主定时任务及实战验证模式。Hal Stack核心组件 🦞
★ 832 📥 212,606
ai-intelligence

ontology

oswalpalash
类型化知识图谱,用于结构化智能体记忆与可组合技能。支持创建/查询实体(人员、项目、任务、事件、文档)及关联...
★ 709 📥 243,384
ai-intelligence

self-improving agent

pskoett
捕获经验教训、错误和纠正,以实现持续改进。使用时机:(1)命令或操作意外失败;(2)用户纠正……
★ 4,054 📥 794,385