Token导航 LogoToken导航

Agent Skill · 研究检索

db-wal-recovery(数据库沃尔玛恢复)

db-wal-recovery 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。

db-wal-recovery 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为权限未标注;许可证为 MIT,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 46,825。

总安装 760周安装 32GitHub Stars 93下载 266核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
MIT

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:db-wal-recovery(数据库沃尔玛恢复)
来源仓库:https://github.com/letta-ai/skills
仓库路径:skills/db-wal-recovery
安装命令:
npx skills add https://github.com/letta-ai/skills --skill db-wal-recovery
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/letta-ai/skills --skill db-wal-recovery

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

db-wal-recovery 指导从损坏或加密的 SQLite WAL 文件中恢复有效数据。核心原则是提取真实存在的数据,而非虚构或猜测内容,避免信息失真。首先验证 journal_mode 是否为 WAL,并定位 main.db、.wal 和 .shm 文件位置。

提供分步诊断流程,但实际恢复效果取决于文件完整性和日志一致性状态。适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/db-wal-recovery

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/letta-ai/skills --skill db-wal-recovery;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    权限没有明确标注,第一次使用时只给一个无关紧要的任务,观察它到底读了哪些文件和网络。

常见问题

6 条

db-wal-recovery 怎么安装?

在终端执行 npx skills add https://github.com/letta-ai/skills --skill db-wal-recovery。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

db-wal-recovery 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

db-wal-recovery 需要什么权限?

权限级别没有明确标注。第一次使用时建议只给无关紧要的任务,观察它实际读写了哪些内容。

db-wal-recovery 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

db-wal-recovery 用的人多吗?

公开数据是累计安装 760 次,近一周安装 32 次,下载 266 次,GitHub 93 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

db-wal-recovery 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
db-wal-recovery

数据库沃尔玛恢复 · 本页

76093权限未标注MIT通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

SQLite WAL Recovery

This skill provides guidance for recovering data from SQLite Write-Ahead Log (WAL) files that appear corrupted, encrypted, or otherwise inaccessible.

Core Principle

Recovery means finding existing data, not generating new data. Never fabricate or guess data based on patterns. The goal is to locate and extract actual data from the source files.

Initial Assessment

Verify WAL Mode and File Presence

  1. Check if the database is in WAL mode: PRAGMA journal_mode;
  2. Identify all database-related files (typically main.db, main.db-wal, main.db-shm)
  3. Critical: If a file appears in one listing method but not another, this is a significant clue requiring investigation, not dismissal

Document Initial Observations

Record all file listings from different tools/methods. Discrepancies between tools often indicate:

  • Special characters in filenames
  • Permission issues
  • Different file encodings
  • Hidden or mounted files
  • Tool-specific behavior differences

Investigation Strategies

When Files Appear to Be Missing

If a WAL file is listed by one tool but inaccessible via another:

  1. Investigate the discrepancy thoroughly - do not dismiss it
  2. Check for special characters: ls -la | cat -A
  3. Examine permissions: ls -la, stat <filename>
  4. Try different access methods: direct path, glob patterns, hex representation
  5. Check if file is a symlink: file <filename>, readlink <filename>
  6. Examine parent directory permissions
  7. Consider if the file might be in a different mount or namespace

When Files Appear Corrupted or Encrypted

Step 1: Examine Raw File Structure
xxd <filename> | head -50
hexdump -C <filename> | head -50
  • Valid SQLite WAL files start with specific magic bytes
  • Look for recognizable patterns vs random-looking data
  • Document what you observe before concluding encryption
Step 2: Systematic Decryption Attempts

If encryption is suspected:

  1. Single-byte XOR: Try common keys (0x00-0xFF)
  2. Multi-byte XOR: Try common patterns, repeating keys
  3. Look for key hints: Check other files in the directory for encryption keys
  4. Examine headers: WAL headers may contain encryption method clues
  5. Try known SQLite encryption libraries: SQLCipher, SEE, etc.
Step 3: Check for Compression

Data may be compressed rather than encrypted:

  • Look for compression signatures (gzip: 1f 8b, zlib: 78 9c, etc.)
  • Try decompression before assuming encryption

When Database Schema Exists but Data is Missing

  1. Query existing data to understand current state
  2. Examine database pages directly with hex tools
  3. Look for data in freelist pages
  4. Check if WAL contains uncommitted transactions

Verification Strategies

Before Concluding Data is Unrecoverable

  • Investigated all tool discrepancies thoroughly
  • Tried multiple access methods for inaccessible files
  • Examined raw bytes of all relevant files
  • Attempted systematic decryption with multiple methods
  • Checked for compression
  • Looked for encryption keys in related files
  • Revisited initial observations for missed clues

Before Submitting Any Solution

  • Solution contains only data actually recovered from files
  • No data was fabricated based on patterns or guesses
  • Can trace each piece of recovered data to its source
  • If data appears to follow a pattern, verified pattern against actual file contents

Common Pitfalls to Avoid

1. Dismissing Tool Discrepancies

Wrong: "The LS tool shows the file but ls doesn't find it, so it doesn't exist."

Right: Investigate why different tools show different results - this often reveals the key to accessing the file.

2. Pattern-Based Data Fabrication

Wrong: "Records 1-5 are apple=100, banana=200... so records 6-10 must be fig=600, grape=700..."

Right: Only include data that can be traced to actual file contents. If data cannot be recovered, report that fact rather than guessing.

3. Reframing "Encrypted" as "Missing"

Wrong: "The WAL is 'encrypted' means the data is missing, so I'll generate likely values."

Right: "Encrypted" means there IS data that needs to be decrypted. Persist with decryption attempts.

4. Premature Abandonment

Wrong: Giving up after simple single-byte XOR doesn't work.

Right: Systematically try multiple decryption approaches before concluding data is unrecoverable.

5. Ignoring Initial Observations

Wrong: Getting stuck and trying new approaches without revisiting earlier clues.

Right: When stuck, return to initial observations - especially any discrepancies or anomalies noted early on.

Recovery Workflow Summary

  1. Document: Record all initial observations and file listings
  2. Verify: Confirm WAL mode and identify all database files
  3. Investigate: Thoroughly explore any discrepancies in file access
  4. Analyze: Examine raw file contents before concluding corruption/encryption
  5. Decrypt: Systematically attempt multiple decryption methods
  6. Validate: Ensure all recovered data traces to actual file contents
  7. Report: Clearly distinguish recovered data from unrecoverable data

When Recovery Fails

If data truly cannot be recovered after exhaustive attempts:

  1. Document all methods attempted
  2. Explain why each method failed
  3. Report what data IS available vs what remains unrecoverable
  4. Never substitute guessed data for unrecovered data

平台分布

Claude Code

29.54%
按下载量换算79

Gemini CLI

24.1%
按下载量换算64

Antigravity

18.81%
按下载量换算50

windsurf

12.35%
按下载量换算33

OpenCode

6.67%
按下载量换算18

Codex

3.29%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

权限未标注

当前来源未能明确判断权限范围,默认进入异常复核队列。

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。

继续浏览同类 Skills