testland/hybrid-search-eval-author
Use when a search system combines keyword (BM25) and vector retrieval, with or without a reranker, and needs numbers: judgment sets, nDCG@k and MRR on fused results, Reciprocal Rank Fusion vs weighted fusion vs single-stage, and whether the reranker earns its latency.
91%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
W011: Third-party content exposure detected (indirect prompt injection risk).
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
The workflow reads external search results and document contents during hybrid search evaluations and relevance testing.
Where we found it
api.anthropic.com
domain · 6 sites
The plugin's LLM-assisted labeling recipe (judgment-sets.md) creates an anthropic.Anthropic() client and calls messages.create() to send query and document text to Anthropic's API for relevance grading, making this a runtime fetch from an external third-party service.
api.cohere.com
domain · 8 sites
The plugin's reranker measurement code (weighted-fusion-and-reranking.md) creates a cohere.ClientV2() and calls co.rerank() to send query text and candidate document contents to Cohere's API for reranking, making this a runtime fetch from an external third-party service.
- Audited
- Security analysis