memory-search-best-practices

记忆系统多刀检索 + recall coverage 策略（8 类题型 recipe）。 Use when: 任务是 "哪些地方提过 X" / "X 的来源 / source map" / "有没有提过 Y / absence check" / "上次到现在变了什么 / delta" / 冷启动 onboard 复杂主题 / 任何召回任务搜了一刀觉得不够。 Not for: 只是选哪个入口走第一刀（用 memory-navigation）/ 已知精确 anchor 单 Read（直接 Read）/ 代码符号查（Grep/LSP）/ 新功能开发（不是 recall 任务）。 Output: 单一 owner 在一条任务链内完成多 query / 多 scope union + coverage matrix（item/source/谁提到/直接 vs 间接）+ 一次终局交付。 GOTCHA: 和 memory-navigation 互补不重叠 — memory-navigation 决定**第一刀走哪个工具**（search vs graph vs list_recent），本 skill 决定**要不要补刀 + 题型对应几刀几路 + 何时停**。Ragdoll家族必加载：opus 系治"我能猜出来 / 碎片够了"停太早病，fable 系治"再确认一轮"停太晚病（双向校准见正文）。

Quality

90%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

An actionable, well-sequenced recall skill body with concrete tool calls and explicit stop criteria. Its main weakness is conciseness — repeated boundary distinctions and inline dated/time-sensitive provenance inflate the token budget.

Suggestions

Consolidate the memory-navigation-vs-this-skill distinction into one section; it currently recurs at lines 8, 132, 140, and 144.

Move dated provenance and spec versions ('2026-05-17 dogfood', '2026-07-08/15 修订', 'v1.2 SW-1') into a single 'Changelog / provenance' or deprecated section so the main body stays evergreen and token-lean.

Hoist the F256 'Related directions'接法 and Ragdoll family telemetry details into a referenced file once they grow, to keep the overview level one-deep.

Dimension	Reasoning	Score
Conciseness	Dense and mostly free of basic-concept padding, but repeats the memory-navigation distinction in four places and carries time-sensitive clutter inline ('2026-05-17 dogfood 实证', '2026-07-08', '2026-07-15 修订', spec version 'v1.2 SW-1') that the rubric penalizes unless isolated in a deprecated section.	3 / 5
Actionability	Fully executable recipes with real calls — 'search_evidence("audhd adhd asd", hybrid, scope=docs, limit=10)', 'graph_resolve(anchor, depth=1, relations=[feature_ref,related_to])', 'get_thread_context', 'list_recent(scope=threads, since=N天)' — and a concrete 5-step worked example.	5 / 5
Workflow Clarity	Multi-step processes are clearly sequenced (8-recipe table, 5-step AUDHD recipe, per-query stop-criteria table) with explicit validation/feedback loops ('≥3 路命中无新 anchor 才停', '高置信命中 → 必 Read 原文', absence '正反两路 + ≥1 个相关概念都 0 命中').	5 / 5
Progressive Disclosure	No bundle files exist; the ~130-line body is organized into clear sections with one-level-deep inline spec links (F200/F256 docs, memory-routing-partial.md). Minor gaps: dated provenance and spec-version pointers are inlined rather than collected in a dedicated reference.	4 / 5
	Total	17 / 20 Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A highly specific, well-triggered description that answers both what and when with concrete operator phrases and explicit boundary guidance. Third-person voice is used throughout with no over-claims.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — '多刀检索 + recall coverage 策略（8 类题型 recipe）', '多 query / 多 scope union + coverage matrix（item/source/谁提到/直接 vs 间接）+ 一次终局交付' — comprehensive coverage of what the skill does.	5 / 5
Completeness	Explicitly answers both what (multi-query recall coverage strategy) and when ('Use when:' with concrete trigger phrases), and adds a 'Not for' boundary clause.	5 / 5
Trigger Term Quality	Natural user phrases inline ('哪些地方提过 X', 'source map', '有没有提过 Y / absence check', '上次到现在变了什么 / delta') plus a dedicated triggers field with 15 synonyms and Chinese/English variants.	5 / 5
Distinctiveness Conflict Risk	Clear niche with an explicit 'Not for' clause distinguishing it from memory-navigation, Read, Grep/LSP, and new-feature work, plus a GOTCHA contrasting memory-navigation — minimal conflict risk.	5 / 5
	Total	20 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	15 / 16 Passed

Repository: zts212653/clowder-ai
Path: cat-cafe-skills/memory-search-best-practices/SKILL.md
Commit: b292744

Reviewed: 6 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.