error-collection

error-collection is a skill for Claude Code, Codex from lfyxhappy/lfcode. It costs 30 tokens per session (1,918 once invoked), scanned A, original, MIT.

A structured record of confirmed engineering mistakes, their root causes, and checks that prevent them from returning. It uses evidence from real application behavior, not just successful builds.

In plain words
What is it for?
It helps document regressions, failed fixes, lifecycle bugs, and the exact checks needed after reloads, restarts, or state changes.
Why use it?
It prevents teams from repeating fixes that only address symptoms or work in one state. It also separates source code success from what users actually see and do.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions OpenCode.

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.

agentmods
npx agentmods add skills/lfyxhappy/lfcode/error-collection
Any agent
npx skills add lfyxhappy/lfcode --skill error-collection
Clone the repo
git clone --depth 1 https://github.com/lfyxhappy/lfcode

Made for: Claude Code, Codex.

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 error-collection

README.md
[![agentmods](https://agentmods.dev/badge/skills/lfyxhappy/lfcode/error-collection.svg)](https://agentmods.dev/skills/lfyxhappy/lfcode/error-collection)
Your own site
<a href="https://agentmods.dev/skills/lfyxhappy/lfcode/error-collection"><img src="https://agentmods.dev/badge/skills/lfyxhappy/lfcode/error-collection.svg" alt="Measured on agentmods" height="20"></a>
Per session 30 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,918 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00030 $0.01918
Opus 5 $0.00015 $0.00959
Sonnet 5 $0.00006 $0.00384
Haiku 4.5 $0.00003 $0.00192

Measured yesterday against content hash 96ee5d146e2c, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

error-collection 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 yesterday.

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.

.lfcode/skills/error-collection/SKILL.md · 90 lines

How it starts

The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.

错误合集

使用时机

当用户指出回归、重复修复失败、运行态与源码不一致,或发现同类低级错误时,先记录事实再继续修复。记录必须基于真实复现,不把“构建成功”当作“功能修复”。

记录格式

每条记录包含:

  • 问题现象和用户复现步骤
  • 运行态证据(页面、DOM、进程、请求结果)
  • 根因和完整生命周期链路
  • 错误决策(为什么之前的修复不成立)
  • 正确修复和明确的非目标
  • 防复发测试、切换/重载/重启验收项

反面案例:供应商用量按钮(2026-08-23)

现象

应用初始页面可以打开“查看用量”,进入具体对话后同一按钮点击无反应;按钮事件返回成功,但浮层节点不存在。切换对话时还可能残留旧供应商状态。

根因

这是一个完整的入口、状态、挂载和验证链路问题,不是单一接口问题:

  • 入口曾经依赖受控 Kobalte Popover。Trigger 自己会 toggle,业务 click handler 又修改同一个 open,造成双重切换;外部 open、Trigger 内部状态和侧栏/会话重渲染叠加后,出现“按钮有点击但浮层不存在”。
  • 通用 Popover 曾加入全局 focusin 失焦关闭。Portal 内容刚挂载、焦点仍在按钮或父级重渲染时,这个监听可能把刚打开的浮层立即关掉。
  • 入口组件一度错误地放进 props.children,导致它不在稳定的 AppLayout action 区域;切换对话、目录或侧栏时组件生命周期不稳定。
  • 能力/供应商状态变化没有一开始就拆成“独立浮层状态 + 当前 provider 数据”,旧会话状态因此可能泄漏到新会话。
  • 验证链路不完整:反复验证首次页面和构建结果,没有在同一预发布运行窗口覆盖进入对话、切换对话、切换供应商、收起再展开侧栏、弹层内点击和弹层外点击。

错误决策与低效原因

  • 把“点击按钮”误读成“按钮外关闭”,交互契约没有先写成可验收的状态机:按钮点击打开/再次点击切换,小窗内部点击保持,小窗外点击关闭,Escape 关闭。
  • 只看源码、类型检查、打包和 HTTP 200,过早把“请求成功”当成“用户看到了弹窗”;没有优先读取真实窗口 DOM、aria-expanded、Portal 节点和当前 provider。
  • 在同一受控 Popover 机制上连续堆 openonClick、微任务和 focus 监听补丁,失败两次后仍未立即切换策略,违反了“同策略重复失败必须换机制”的纪律。
  • 没有先画完整生命周期:入口挂载位置、provider 来源、状态所有权、Portal 挂载点、外部点击边界、会话/侧栏重渲染边界;因此修复了局部症状又引入新的生命周期问题。
  • 没有把每个失败尝试按假设、证据、策略和回归项记录下来,导致重复排查、重复打包和重复人工确认,效率显著下降。
  • 运行态验证没有一次性覆盖完整矩阵,迫使用户多轮发现“进入对话后失效”“切换对话后失效”“收起再展开才恢复”等本应由自动化验收捕获的问题。

最终修复与边界

  • ProviderQuotaSidebarAction 使用自己的显式 open 状态和原生事件监听,不再依赖 Kobalte Trigger 的内部 toggle。
  • 小窗通过 Portal mount={document.body} 独立挂载,使用固定定位和 z-index: 2147483647,不受侧栏 overflow、层叠上下文和父节点裁剪影响。
  • 全局监听只处理 pointerdown 和 Escape:目标在触发按钮或小窗内部时保持打开,只有小窗外才关闭;小窗关闭按钮仍可显式关闭。
  • 额度请求按 providerID 使用独立 Map 缓存。当前对话只提供当前 provider 的 ID/名称,不能拥有或控制小窗生命周期。
  • 入口固定通过 AppLayout quotaAction 渲染在稳定 action 区域;provider 变化只更新显示数据,不把对话状态写入小窗。

防复发规则

  • 共享 Popover 组件必须验证:首次打开、关闭、连续开关、切换路由/会话、重新挂载。
  • 受控 Trigger 若有内部 toggle,不得在同一同步 click handler 中再次改变相同状态。
  • 会话或供应商变化时,使用当前 provider key 重建有状态的浮层实例。
  • 复杂组件连续两次同策略失败后,必须改用独立显式状态和显式渲染,不再继续堆补丁。
  • 验收必须查询真实运行窗口中的 aria-expanded、浮层 DOM、data-provider-id,并确认当前对话供应商一致。
  • 每次涉及浮层、侧栏或会话路由的修改,必须在预发布安装副本完成以下矩阵:首次打开、再次点击关闭、小窗内部点击保持、小窗外点击关闭、Escape 关闭、切换对话、切换供应商、收起/展开侧栏、刷新/重新挂载。
  • 必须记录实际运行进程、窗口 ID、DOM 快照和关键属性;typecheck、构建、接口 200 只能作为前置证据,不能作为交付证据。
  • 连续两次同一交互机制失败后,停止补丁式修改,重新梳理状态所有权和 DOM 挂载边界,并改用可区分根因的实验。
  • 交付前先写 bug contract 和验收矩阵,再实现;每次失败都记录 hypothesis、prediction、strategy-key、verification 和 regressions,避免重复劳动。

Read the full file on GitHub · 90 lines

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. yesterday First seen · 90 lines · 30 tokens per session scan A 96ee5d146e2c

Subscribe to this mod's changes

error-collection is a skill published in the GitHub repository lfyxhappy/lfcode (2 stars, last pushed 3d ago), licensed MIT. It adds 30 tokens to every session and 1,918 once invoked, about $0.0002 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.

Related

Other skills, from other repositories

bug-fix

半自动修复 bug 的标准作业流程:复现 → 定位根因(要证据)→ plan 出方案 → 人工审批 → 实现 → 测试验证。专为"为 AI 代码兜底"的修复闭环设计。.

rushengzhou/sid-code · 62 tokens

mandu-mcp-verify

편집 후 검증 루프. 파일 편집 직후, "check 해줘", stop-hook 자동 호출. ateautopipeline + guardcheck + doctor 를 병렬로. 개별 ate 도구로 수동 분해 금지 — 실패했을 때만 drill-down.

konamgil/mandu · 66 tokens

termcanvas-agent-sync-debugging

Diagnose TermCanvas agentmux nodes that fail to appear and terminal canvases that become janky, resize repeatedly, or show corrupted/unreadable xterm text as terminal count grows, plus duplicate terminal views created after a release relaunch. Use when a child agent is missing, restored terminals duplicate or move to…

lout33/termcanvas · 116 tokens

repair-prevention-assessment

修复防复发完成门禁 Owner — 任何 fix、self-fix 或 dev 中的 repair 切片在 accepted 前使用;形成可机器验证的 RepairPreventionAssessmentV1,分离当前关闭证据与长期前瞻有效性,避免 active 工作流依赖未部署的 gray 能力。.

devcodex-labs/devcodex · 75 tokens

ci-self-heal

针对 CI 失败日志输出结构化诊断与修复建议. 识别失败分类(test/lint/build/type/dependency/config/flaky/timeout) / 根因假设 / 可执行 fix, 引用具体 file:line. 专为 PR-to-Prod 流程的 CI 卡点兜底设计.

rushengzhou/sid-code · 73 tokens

Check work (verify against criteria)

Verify an implementation against acceptance criteria with a reviewer and a tester.

AuraVixStudio/caelo · 20 tokens