iterative-retrieval

为了解决子智能体（Sub-agent）的上下文问题，逐步优化上下文获取的模式

Quality

48%

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

Fix and improve this skill with Tessl

tessl review fix ./docs/ja-JP/skills/iterative-retrieval/SKILL.md

Quality

Content

62%

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

The content presents a clear, well-sequenced iterative-retrieval workflow with explicit checkpoints and good worked examples, but it leans on pseudocode-style helper stubs and carries a decorative diagram plus concept explanation that dilute conciseness. No bundle files are used, so all detail lives inline.

Suggestions

Make code executable by defining or inlining the helper stubs (retrieveFiles, scoreRelevance, extractPatterns, mergeContext, hasCriticalGaps) or explicitly noting them as conceptual placeholders to implement.

Trim the ASCII diagram and the '问题背景' section, which restate concepts Claude already knows, to recover token budget.

Consider moving the two worked cases or the full code listings into a referenced file (e.g., EXAMPLES.md) to better separate overview from detail.

Dimension	Reasoning	Score
Conciseness	Most of the body is actionable (code, scoring criteria, cases), but the ASCII dispatch diagram and the '问题背景' section explain a concept Claude already knows and add length without earning their tokens, fitting 'mostly efficient but includes some unnecessary explanation'.	2 / 3
Actionability	It gives concrete structure and examples, but the code calls undefined helpers (retrieveFiles, scoreRelevance, extractPatterns, mergeContext, hasCriticalGaps), making it pseudocode rather than copy-paste executable, matching 'some concrete guidance but incomplete; pseudocode instead of executable code'.	2 / 3
Workflow Clarity	The 4-phase DISPATCH→EVALUATE→REFINE→LOOP cycle is clearly sequenced with explicit validation checkpoints (relevance >= 0.7 threshold, highRelevance count check, hasCriticalGaps) and a max-3-cycle feedback loop, matching 'clear sequence with explicit validation steps; feedback loops'.	3 / 3
Progressive Disclosure	It is well-organized into sections, but at ~210 lines everything is inline in one file with no split-out reference files (only external links), fitting 'some structure but content that should be separate is inline'; the under-50-line shortcut does not apply.	2 / 3
	Total	9 / 12 Passed

Description

35%

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

The description identifies a real, fairly specific niche but is written in technical jargon with no natural trigger terms and no 'Use when...' guidance, capping both completeness and trigger quality. It is distinguishable but not crisply so.

Suggestions

Add an explicit 'Use when...' clause stating when to activate (e.g., 'Use when spawning subagents whose needed codebase context cannot be predicted upfront, or when agent tasks fail with missing/too-large context').

Replace jargon with natural user-facing keywords (e.g., 'subagent context', 'missing context', 'too much context', 'codebase retrieval') so users would naturally say them.

Lead with concrete actions instead of 'a pattern' (e.g., 'Iteratively retrieves, scores, and refines codebase context for subagents').

Dimension	Reasoning	Score
Specificity	Names the domain ('上下文检索', '子智能体上下文问题') and a couple of actions ('逐步优化', '解决'), but the actions are abstract ('a pattern') rather than concrete verbs like extract/fill/merge, so it stops at 'names domain and some actions, not comprehensive'.	2 / 3
Completeness	It states what the skill does (a pattern for iteratively optimizing context retrieval) but gives no explicit 'when' trigger guidance, so per the guideline a missing 'Use when...' clause caps completeness at 2.	2 / 3
Trigger Term Quality	The phrasing is technical jargon ('迭代检索模式', '上下文问题', '子智能体') with no natural keywords a user would actually say; there is no 'Use when...' clause, matching the 'no natural keywords; technical jargon' anchor.	1 / 3
Distinctiveness Conflict Risk	The subagent-context-retrieval niche is fairly specific, but without explicit triggers it could still overlap with general RAG or retrieval skills, fitting 'somewhat specific but could still overlap'.	2 / 3
	Total	7 / 12 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: xu-xiang/everything-claude-code-zh
Path: docs/ja-JP/skills/iterative-retrieval/SKILL.md
Commit: dfbf946

Reviewed: 2 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.