Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/hyh0620/mcp-knowledge-servicenpx agentmods add skills/hyh0620/mcp-knowledge-service/evaluate-retrievalWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/hyh0620/mcp-knowledge-service/evaluate-retrieval)<a href="https://agentmods.dev/skills/hyh0620/mcp-knowledge-service/evaluate-retrieval"><img src="https://agentmods.dev/badge/skills/hyh0620/mcp-knowledge-service/evaluate-retrieval/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/hyh0620/mcp-knowledge-service/evaluate-retrieval"><img src="https://agentmods.dev/badge/skills/hyh0620/mcp-knowledge-service/evaluate-retrieval.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00018 | $0.00175 |
| Opus 5 | $0.00009 | $0.00088 |
| Sonnet 5 | $0.00004 | $0.00035 |
| Haiku 4.5 | $0.00002 | $0.00017 |
Grade A, and why
evaluate-retrieval scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Evaluate Retrieval
Inputs
- Collection name.
- Golden test set with query and expected source.
Pipeline
- Confirm collection exists:
python scripts/query.py --query "health check" --collection <COLLECTION> --top-k 1 - Run retrieval cases using the project evaluation script or a small client script.
- For each case record:
- returned sources
- first relevant rank
- citation count
- latency
- Compute:
- Hit@1
- Hit@3
- MRR
- citation expected-source match
Rules
- Report numerator and denominator.
- Do not use LLM subjective scoring as retrieval accuracy.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 35 lines · 18 tokens per session scan A 0f92a54f51e7
evaluate-retrieval is a skill published in the GitHub repository hyh0620/mcp-knowledge-service (0 stars, last pushed 1mo ago), licensed MIT. It adds 18 tokens to every session and 175 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-01.
Other skills, from other repositories
vector-hybrid-search
Retrieve knowledge from a vector-store collection via the vector-mcp MCP server's vectorsearch tool — semantic (ANN) search, lexical BM25 search, or a hybrid of the two fused with Reciprocal Rank Fusion. Use when the agent must answer a question from an indexed corpus, pull top-k relevant chunks for RAG context, or…
using-agent-brain
Expert Agent Brain skill for document search with BM25 keyword, semantic vector, hybrid, graph, and multi retrieval modes. Use when asked to "search documentation", "query domain", "find in docs", "bm25 search", "hybrid search", "semantic search", "graph search", "multi search", "find dependencies", "code…
vector-backend-operations
Select and connect the right vector-store backend for the vector-mcp MCP server — chromadb, postgres/pgvector, qdrant, couchbase, or mongodb — and supply the correct dbtype/connection parameters that every collection and search call needs. Use when the agent must decide which engine to target, wire up…
vector-collection-management
Create, populate, list, and delete vector-store collections through the vector-mcp MCP server's vectorcollectionmanagement tool. Use when the agent must stand up a new RAG collection, ingest documents (from a directory, file paths/URLs, or raw text) into an existing collection, enumerate collections, or drop one …
eval-set
Build and run a project-specific retrieval eval so changes to your rules or base prompt are scored, not eyeballed. Mirrors tests/groundtruth.json + clawness eval: you write prompt→expected-rule cases, then measure MRR@k and hit-rate before and after an edit. Run it after trimming a base prompt into ranked retrieval…
Hybrid Search Architect
Designs a hybrid retrieval pipeline combining dense vector search and BM25 sparse search with reciprocal rank fusion, and explains when to use each configuration.