rag-assistant

Knowledge base assistant that retrieves and cites documents from a curated index. Use when answering questions that must be grounded in specific source material with citations.

Quality

78%

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 ./modules/skills/src/main/resources/META-INF/skills/rag-assistant/SKILL.md

Quality

Content

78%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.

A compact, well-structured instruction skill for grounded Q&A over a curated knowledge base: the workflow is sequenced, guardrails are explicit, and nothing pads the token budget beyond some duplication. Its gaps are minor — a citation format example, a recovery path for empty search results, and redundant tool/skill listings.

Suggestions

Remove the duplicate tool listing — keep either the intro bullets or the '## Tools' section, not both — and drop the vague '## Skills' section to tighten conciseness.

Add a one-line example of the expected citation format so answers are consistently attributable.

Add a feedback step for empty or weak search results, e.g., 'If a search returns nothing, try different keywords or run list_sources to see what's available before concluding the answer isn't in the knowledge base.'

Dimension	Reasoning	Score
Conciseness	The body is lean overall with no concept explanations Claude doesn't need, but the tool list is stated twice (the intro bullet list and the '## Tools' section) and the vague '## Skills' section ('Explain transports, AI modules, agents...') is padding that could be trimmed. This matches the anchor of efficient content with minor over-explanation, short of the every-token-earns-its-place level.	4 / 5
Actionability	The numbered workflow gives concrete, executable behavioral guidance ('Compose your answer using ONLY information from the documents', 'Cite which document(s)', 'say so honestly'), and per the scoring notes an instruction-only skill is not penalized for lacking code. It falls short of fully executable because no citation format example is given and tool invocation is only described, not demonstrated.	4 / 5
Workflow Clarity	The 5-step sequence is clearly ordered and ends with an explicit honesty checkpoint when documents are insufficient, which serves as a validation step. A 5 would require fuller feedback loops — e.g., what to do when a search returns nothing (retry with different keywords, list sources first) — which are absent.	4 / 5
Progressive Disclosure	The skill is under 50 lines with no bundle files (no references/, scripts/, or assets/ directories exist), and per the rubric's own guidance a short skill with well-organized sections (Skills, Tools, Guardrails) and no need for external references scores 5. Sections are clearly headed and easy to navigate.	5 / 5
	Total	17 / 20 Passed

Description

78%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 concise, well-formed description that clearly states both capability and an explicit 'Use when' trigger in third person. Its main limitation is breadth: it names only two concrete actions (retrieve, cite) and omits synonyms like RAG or source-document phrasing that users might naturally say.

Suggestions

Add one or two more concrete capabilities to the what-clause (e.g., 'searches by topic or keyword, compares information across sources') to lift specificity toward comprehensive coverage.

Include natural synonyms users would say, such as 'RAG', 'answer from the docs', or 'source-grounded answers', to broaden trigger-term coverage.

Dimension	Reasoning	Score
Specificity	The description names the domain and two concrete actions — 'retrieves and cites documents from a curated index' — but stops there, matching the anchor for 1-2 concrete actions without comprehensive coverage. A score of 4 would require several specific actions (e.g., searching, summarizing, comparing sources), and 2 would require only a generic domain mention.	3 / 5
Completeness	Both parts are explicitly answered: the what ('retrieves and cites documents from a curated index') and a clearly stated when ('Use when answering questions that must be grounded in specific source material with citations') with concrete trigger phrasing. This matches the top anchor; a 4 would apply only if the when-clause were weakly implied or less specific.	5 / 5
Trigger Term Quality	Natural terms like 'knowledge base', 'documents', 'citations', 'grounded in specific source material' cover the phrasing a user would plausibly say when needing this skill. It misses common synonyms and variations (e.g., 'RAG', 'answer from my docs', 'source documents'), so it falls just short of the comprehensive synonym coverage of a 5.	4 / 5
Distinctiveness Conflict Risk	The curated-index/citation grounding niche is mostly distinct from generic document or search skills, with clear triggers reducing wrong-skill activation. Minor overlap risk remains with closely related document-QA or research skills, which is exactly the anchor-4 condition rather than the fully distinct niche of a 5.	4 / 5
	Total	16 / 20 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning

	Total	14 / 16 Passed

Repository: Atmosphere/atmosphere
Path: modules/skills/src/main/resources/META-INF/skills/rag-assistant/SKILL.md
Commit: 1c968d5

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.