CtrlK
BlogDocsLog inGet started
Tessl Logo

iterative-retrieval

Pattern for progressively refining context retrieval to solve the subagent context problem

38

Quality

35%

Does it follow best practices?

Impact

No eval scenarios have been run

SecuritybySnyk

Passed

No known issues

Optimize this skill with Tessl

npx tessl skill review --optimize ./docs/zh-TW/skills/iterative-retrieval/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

62%

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

This skill clearly articulates a useful pattern for iterative context retrieval with good workflow structure and practical examples. However, it suffers from moderate verbosity in the problem description section (explaining things Claude already knows), and the code examples are illustrative pseudocode rather than truly executable implementations. The progressive disclosure is weak given the document length and lack of supporting bundle files.

Suggestions

Trim the 'Problem' section significantly — Claude already understands subagent context limitations; a single sentence framing is sufficient.

Make code examples more actionable by either providing real implementations of helper functions (scoreRelevance, identifyGaps) or replacing with actual tool-use patterns (e.g., grep, find, ripgrep commands) that Claude can execute.

Replace the Twitter/X link with stable documentation or inline the key coordination concepts, as social media links are ephemeral.

DimensionReasoningScore

Conciseness

The skill includes some unnecessary explanation of the problem space (Claude already understands context limitations of subagents) and the ASCII diagram adds visual bulk. However, the code examples and practical examples are reasonably efficient. The problem/solution framing could be much tighter.

2 / 3

Actionability

The code examples illustrate the pattern but are pseudocode-like (functions like scoreRelevance, explainRelevance, identifyGaps are undefined placeholders). The practical examples in text format are helpful but not executable. The agent integration section provides a concrete prompt template, which is useful but brief.

2 / 3

Workflow Clarity

The 4-stage cycle is clearly sequenced (DISPATCH → EVALUATE → REFINE → LOOP) with an explicit termination condition (3 cycles max, or sufficient high-relevance files). The loop includes validation checkpoints (relevance scoring, gap identification) and a clear feedback loop for refinement. The two practical examples demonstrate the workflow concretely.

3 / 3

Progressive Disclosure

The content is a single monolithic file with no bundle files. The 'Related' section references external resources but one is a Twitter/X link (potentially ephemeral) and the other references are vague ('continuous-learning skill', agent definitions directory). For a skill of this length (~150 lines), some content like the detailed code examples could be split into referenced files.

2 / 3

Total

9

/

12

Passed

Description

7%

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

This description is too abstract and jargon-heavy to be effective for skill selection. It reads more like an academic concept label than an actionable skill description, lacking concrete actions, natural trigger terms, and any explicit 'use when' guidance.

Suggestions

Replace abstract language with concrete actions, e.g., 'Iteratively narrows search queries to retrieve relevant context for sub-agents, using progressive filtering and re-ranking of results.'

Add an explicit 'Use when...' clause with natural trigger terms, e.g., 'Use when a subagent needs to find relevant context from a large codebase, knowledge base, or document set and initial retrieval results are insufficient.'

Include user-facing keywords that someone would naturally use, such as 'context retrieval', 'search refinement', 'RAG', 'agent context', 'knowledge lookup'.

DimensionReasoningScore

Specificity

The description uses abstract language like 'progressively refining context retrieval' and 'subagent context problem' without listing any concrete actions. It describes a pattern concept rather than specific capabilities.

1 / 3

Completeness

The description vaguely addresses 'what' (a pattern for refining context retrieval) but provides no 'when' clause or explicit trigger guidance. Both dimensions are weak.

1 / 3

Trigger Term Quality

The terms used are highly technical jargon ('subagent context problem', 'progressively refining context retrieval') that users would not naturally say. No common user-facing keywords are included.

1 / 3

Distinctiveness Conflict Risk

The phrase 'subagent context problem' is somewhat niche and specific to a particular architectural concern, which reduces conflict risk slightly, but the overall vagueness could still cause confusion with other retrieval or agent-related skills.

2 / 3

Total

5

/

12

Passed

Validation

100%

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

Validation11 / 11 Passed

Validation for skill structure

No warnings or errors.

Repository
affaan-m/everything-claude-code
Reviewed

Table of Contents

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.