ai-product

Every product will be AI-powered. The question is whether you'll build it right or ship a demo that falls apart in production.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Passed

No known issues

Quality

Content

50%

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

The content is a well-organized, code-rich catalog of AI product patterns and pitfalls that is broadly actionable, but it is verbose with concept explanations Claude already knows, contains pseudocode relying on undefined helpers, and is a monolithic file that would benefit from progressive disclosure into separate reference files.

Suggestions

Trim 'Why this breaks' and 'Situation' narrative prose that restates concepts Claude already knows (probabilistic LLMs, finite context windows) to improve token efficiency.

Replace undefined helper functions (callLLM, rerank, embed, CircuitBreaker) and invalid syntax with complete, executable code or clearly justified stubs.

Split the large Validation Checks, Sharp Edges, and Collaboration sections into separate reference files linked from a concise SKILL.md overview to improve progressive disclosure.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient with concrete code, but it explains concepts Claude already knows (e.g., 'LLMs are probabilistic', 'Context windows are finite', 'LLMs hallucinate') and pads each sharp edge with narrative 'Situation/Why this breaks' prose that could be trimmed.	2 / 3
Actionability	Many sections give concrete, near-executable TypeScript (zod validation, OpenAI streaming, caching), but numerous examples rely on undefined helpers (callLLM, CircuitBreaker, rerank, fullTextSearch, embed, verifySourceExists, generateWithSources) and invalid syntax like vectorDB.search(embedding, topK: 20), making them illustrative rather than copy-paste ready.	2 / 3
Workflow Clarity	Ordered sequences appear (e.g., the RAG hybrid search steps, the 'Try in order: Better prompts -> RAG -> Fine-tuning' ladder, and the Collaboration workflows), but no validation checkpoints or feedback loops are present for these operations, which caps clarity at 2.	2 / 3
Progressive Disclosure	The body is well-sectioned with clear headers (Principles, Patterns, Sharp Edges, Validation Checks, Collaboration), but no bundle files exist and the ~750-line document is monolithic with content (validation checks, sharp edges, collaboration) that should be split into separate referenced files.	2 / 3
	Total	8 / 12 Passed

Description

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 frontmatter description is a marketing tagline rather than a capability statement: it names no concrete actions, omits trigger terms, and provides no "use when" guidance. It fails to tell Claude what the skill does or when to invoke it.

Suggestions

Replace the tagline with a concrete capability statement listing specific actions (e.g., 'Design LLM integrations, RAG architectures, prompt versioning, streaming, and cost optimization').

Add an explicit 'Use when...' trigger clause naming natural terms users would say (LLM, RAG, chatbot, prompt engineering, AI feature).

Rewrite in third person and remove the second-person rhetorical question to improve specificity.

Dimension	Reasoning	Score
Specificity	The description is a rhetorical tagline ("Every product will be AI-powered. The question is whether you'll build it right or ship a demo that falls apart in production.") naming no concrete actions; it is vague rather than listing specific capabilities like LLM integration, RAG, or prompt engineering. The second-person phrasing ("you'll build it") would normally reduce specificity by 1, but it is already at the scale minimum.	1 / 3
Completeness	It answers neither "what does this do" concretely nor "when should Claude use it"; there is no "Use when..." clause and the capability description is absent.	1 / 3
Trigger Term Quality	No natural trigger terms a user would say appear; there is no mention of "AI product", "LLM", "RAG", "prompt", or "chatbot", only abstract marketing language.	1 / 3
Distinctiveness Conflict Risk	The tagline is extremely generic and would not distinguish this skill from a general software-engineering or product skill, creating high conflict risk.	1 / 3
	Total	4 / 12 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
skill_md_line_count	SKILL.md is long (754 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: sickn33/antigravity-awesome-skills
Commit: 9fc5f98

Reviewed: about 3 hours 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.