AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Agentic Retrieval

skill-lihongwei-cn-lihongwei-cn-agentic-retrieval · by LiHongwei-cn

Agentic 检索思维 — 蒙多的记忆不只是搜索,是主动推理

No reviews yet
0 installs
31 views
0.0% view→install

Install

$ agentstack add skill-lihongwei-cn-lihongwei-cn-agentic-retrieval

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-lihongwei-cn-lihongwei-cn-agentic-retrieval)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Agentic Retrieval? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Agentic 检索思维 Skill

核心思想

传统 RAG:Query → Retrieve → Generate Agentic RAG:Query → 自问规划 → Retrieve → 验证 → Synthesize → Generate

蒙多的检索不是被动搜索,是主动推理。

检索决策树

面对一个问题时,蒙多先问自己:

1. 我现在知道答案吗?
   ├─ 确定知道 → 直接回答
   ├─ 不确定 → 需要检索
   └─ 完全不知道 → 必须检索

2. 如果需要检索,去哪找?
   ├─ 本地记忆(memory.db)→ recall/search
   ├─ 项目文件(当前目录)→ search_files
   ├─ 对话历史(conversations)→ FTS5
   ├─ 外部知识(互联网)→ web_search
   └─ 代码库(github等)→ http_request

3. 检索到了,可信吗?
   ├─ 多源一致 → 高可信
   ├─ 单源 → 需要交叉验证
   └─ 矛盾 → 重新检索或标注不确定性

自问清单(Self-Questioning)

在回答复杂问题前,蒙多必须过一遍:

  1. 这个问题的前提成立吗? — 用户的假设可能有误
  2. 我有直接证据吗? — 还是在靠推测?
  3. 有没有遗漏的关键信息? — 信息缺口在哪?
  4. 我的结论能被证伪吗? — 如果不能,可能是废话
  5. 最可能出错的地方是什么? — 主动找薄弱点

检索策略

策略一:关键词拆解

  • 从问题中提取核心关键词
  • 用不同关键词组合多次检索
  • 合并去重

策略二:时间线追踪

  • 先找最近的记录(新鲜度优先)
  • 再找历史记录(完整性)
  • 对比变化

策略三:上下文扩展

  • 找到一个线索后,沿着关联扩展
  • 同一项目的其他文件
  • 同一类别的其他记忆

策略四:反向验证

  • 找到答案后,反过来搜索"反面证据"
  • 如果找到反面证据,标注不确定性

与蒙多记忆系统的集成

蒙多 memory.py 已有:

  • FTS5 全文检索(对话历史)
  • LIKE 后备(中文支持)
  • 自动提取(规则模式)
  • 项目隔离(按目录过滤)

需要增强的:

  1. 自问层 — 检索前先判断是否真的需要检索
  2. 多源融合 — 同时查记忆、文件、网络,合并结果
  3. 可信度标注 — 每条检索结果标注可信度
  4. 检索链路记录 — 记录检索过程,用于复盘

实践示例

用户问:"蒙多的策略引擎支持几级优先级?"

蒙多的 Agentic 检索过程:
1. 自问:我知道 policy.py 的结构吗?→ 不完全确定
2. 规划:先查 policy.py 的 Severity 枚举 → 再查 PolicyEngine 的 evaluate
3. 检索:search_files("Severity|priority", path="policy.py")
4. 结果:Severity 有 LOW(1)/MEDIUM(2)/HIGH(3)/CRITICAL(4) 四级
5. 验证:确认 PolicyEngine.evaluate 确实用了这个排序
6. 回答:四级优先级,从 LOW(1) 到 CRITICAL(4)

参考资料

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.