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.
npx skills add dslsdzc/rev-skills --skill re-feedbackgit clone --depth 1 https://github.com/dslsdzc/rev-skillsWrote 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.
[](https://agentmods.dev/skills/dslsdzc/rev-skills/re-feedback)<a href="https://agentmods.dev/skills/dslsdzc/rev-skills/re-feedback"><img src="https://agentmods.dev/badge/skills/dslsdzc/rev-skills/re-feedback/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.
<a href="https://agentmods.dev/skills/dslsdzc/rev-skills/re-feedback"><img src="https://agentmods.dev/badge/skills/dslsdzc/rev-skills/re-feedback.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00082 | $0.01763 |
| Opus 5 | $0.00041 | $0.00881 |
| Sonnet 5 | $0.00016 | $0.00353 |
| Haiku 4.5 | $0.00008 | $0.00176 |
Grade A, and why
re-feedback 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 9d ago.
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.
How it starts
The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
经验反馈
完整工作流
- 收集（三源归一）：
- 会话复盘：本次分析会话的踩坑/新方法（[[re-analyze]] 第四步调用时已就绪）
- 文章扫描：
bin/wxsource.mjs抓取看雪/微信文章（kanxue thread <ID> --md/wechat <URL> --md），逐篇蒸馏 - 手动输入：用户一句话/一段笔记
- 蒸馏与脱敏：按
references/intake.md—— 坑格式**标题**：现象——…；原因——…；对策——…，同步脱敏 - 归域分类：
references/intake.md决策表 → grep 技能库兜底 → 仍不确定问用户（不猜） - 三档处理（中性呈现，不带推荐标注）：
先问用户档位：「这次经验怎么处理？1. 发表 issue 2. 本地入库 3. 不入库」，选定后再执行对应档，不预设默认档
- ① 发表 issue：
which gh+gh auth status探测；可用则gh issue create --repo dslsdzc/rev-skills --label 经验，标题「经验: <技能>: <一句话>」，正文按references/issue-template.md；不可用 → 输出 markdown 草稿供手动提交；label 报不存在 → 去掉 --label 重试或提示用户先创建；仓库不可达 → 输出草稿 - ② 本地入库：grep 去重 → 追加到该技能
references/experience.md（无则创建，条目含来源+日期）→ SKILL.md「常见坑与陷阱」保留精选并追加 experience.md 链接（先建文件再加链接，避免断链）→npm test校验 → 按步骤 5 约定提交（或先询问用户） - ③ 不入库：确认后结束
- ① 发表 issue：
- （文章扫描路径可选）提交 commit：「经验: 第N轮扫描入库X篇——<篇目>」，message 列出每技能新增条目
何时走哪条路（选择树）
- 输入 = 本次分析会话摘要 → 会话复盘源；目标 = 沉淀本次实战
- 输入 = 文章 URL / 帖子 ID / 文章文本 → 文章扫描源；逐篇蒸馏，按 intake.md 标准跳过工具/推广帖
- 输入 = 用户一句话/笔记 → 手动源；直接蒸馏
- 三档选择（中性呈现）：1. 发表 issue（可公开分享、他人可复用）2. 本地入库（本库沉淀、随库发布）3. 不入库（仅会话内保留）
跨域联合
- [[re-analyze]] 第四步（可选）调用本网关做会话复盘
- 归域目标为全部技能（[[re-binary-core]] [[re-malware]] [[re-firmware]] [[re-protocol]] [[re-mobile]] [[re-anti-analysis]] [[re-cracking]] [[re-vuln]] [[re-ctf]] [[re-managed]] [[re-forensics]] 及其原子技能）
- 文章抓取用仓库
bin/wxsource.mjs；入库后用npm test（validate.mjs）校验结构
常见坑与陷阱
- 蒸馏不脱敏直接入库/发 issue → 泄漏项目身份 —— 红线 2，入库前逐项列专有名词自检
- 呈现带「最推荐」等强推 → 绑架用户判断 —— 红线 1，只中性列选项
- 归域不确定硬猜 → 经验写错技能、下次检索不到 —— 问用户
- 入库后不跑 npm test → 断链/格式错破坏 CI —— 每次入库后必须
npm test - 重复条目重复入库 → 库膨胀 —— 入库前 grep 关键词去重
- 文章扫描不跳过工具/推广帖 → 经验贬值 —— 按 intake.md 跳过标准过滤
- 信任工具输出前先断言前置条件：现象——工具运行看似正常但产出空/宽松/无操作（缺官方运行时返回空输出、架构不符时 hook 零命中、带壳重打包真机黑屏、屏幕未唤醒截图空白）；原因——工具都有硬前置条件（官方运行时/原生架构/未加壳/环境就绪），违反时不报错而安静失败；对策——采信任何输出前显式断言其前置条件在该目标上成立，静默与空结果一律先当可疑；三条结构互异的直路均失败即被运行时不变量封锁，停止重试，改换环境/抽象层/目标（来源：reverse-skills（inliver233），MIT）
- 多代理协作状态写入交接文档：现象——多代理并行分析对账不上、死路重走、假设反复验证；原因——状态只存在聊天历史，各代理对「已确认/已排除」无共同基准；对策——单一控制器作唯一验证权威（代理假设须经控制器装机复现才算确认），每代理每轮一个目标，操作现场/已确认事实（带验证方式）/必填的已排除清单/建议第一步写入交接文档，对账按证据类别排序（运行时 > 静态 > 假设）（来源：reverse-skills（inliver233），MIT）
- 蒸馏措辞触发协作模型拒答：现象——含「破解/绕过」等词的蒸馏产物在协作流程中遭拒答；原因——触发词被内容策略拦截，与内容质量无关；对策——蒸馏阶段系统性改写为工程语言（代码恢复/兼容性适配/完整性校验适配/运行时签名适配），并随附授权前提声明（自有代码恢复/互操作/安全审计/CTF/教育，不授权对抗受保护软件），与红线 1 呈现中性同源，入库前照此自检（来源：reverse-skills（inliver233），MIT）
- 更多经验见本技能 [[experience]]（经验全集）
What ships with it
3 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.
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.
- 9d ago First seen · 58 lines · 82 tokens per session scan A 2d9e14d026c8
re-feedback is a skill published in the GitHub repository dslsdzc/rev-skills (50 stars, last pushed 11d ago), licensed Apache-2.0. It adds 82 tokens to every session and 1,763 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
analyzing-golang-malware-with-ghidra
Reverse engineer Go-compiled malware in Ghidra by parsing Go buildinfo and pclntab structures, recovering stripped/obfuscated function names (e.g. via GoResolver), and extracting embedded module/dependency strings and types from Go binaries. Use when analyzing a Go-language malware sample, deobfuscating a…
analyzing-golang-malware-with-ghidra
Reverse engineer Go-compiled malware in Ghidra by parsing Go buildinfo and pclntab structures, recovering stripped/obfuscated function names (e.g. via GoResolver), and extracting embedded module/dependency strings and types from Go binaries. Use when analyzing a Go-language malware sample, deobfuscating a…
Reverse Engineering & Binary Analysis
Binary analysis, assembly interpretation, disassembly, decompilation, firmware RE, and protocol reverse engineering.
analyzing-golang-malware-with-ghidra
Reverse engineer Go-compiled malware using Ghidra with specialized scripts for function recovery, string extraction, and type reconstruction in stripped Go binaries.
compact
Use when context is running long during a pentest engagement, at phase boundaries after completing a full phase, or when the user says "compact", "compress context", "trim context", or "save tokens". Compresses completed phase outputs while keeping session.json as the authoritative source of truth.
reverse-engineering-arm-binaries
Reverse engineers ARM/AArch64 malware by identifying the architecture and instruction set state (ARM/Thumb), parsing ELF/Mach-O ARM headers, and orienting analysis around the ARM calling convention. Activates for requests to reverse ARM binaries, analyze AArch64 malware, or handle ARM/Thumb instruction-set decoding.