retrieval-rag skillA
retrieval-rag is agent-read markdown (skill) from nahid-sparktales/agent-dispatcher: Build and fix retrieval that actually returns the right passage — structure-aware chunking, one pinned embedding model, lexical plus vector search fused, reranking, and a recall measurement that is run separately from the generator. Use when a RAG system answers wrong or vaguely, when an index is being designed or reindexed, or when someone proposes a prompt change to fix what is really a retrieval miss. Not for prompt or output-quality work once the right passage is already in context, not for .
Indexed from public GitHub and served as immutable, content-addressed versions. Install it pinned to an exact SHA-256 with the mdr CLI, and every file is verified against the hash recorded here before it reaches your agent. The deterministic audit below grades the latest version, and the same file always earns the same grade.
What the file says
# Retrieval for RAG Most "the model hallucinated" reports are retrieval failures wearing a generation costume. The gold passage was never in the context window, and no prompt edit can fix that. The first job is always to find out which half is broken. ## When this fires A system retrieves documents and feeds them to a model, and the answers are wrong, thin, or unsourced — or an index is being built or rebuilt. It does not fire when the correct passage is demonstrably in context and the model still answers badly; that is a generation problem. ## Procedure 1. **Split the failure before fixing anything.** Take 10 real failing questions. For each, check whether the passage that answers it appears in the retrieved set at all. Retrieved and answered wrong is a generation problem; never retrieved is a retrieval problem. Do not proceed on a guess — this one check redirects most of the work. 2. **Build the evaluation set before touching the index.** 30–50 real questions, each labelled with the id(s) of the passage that answers it. Get them from real logs where possible. Without this set every later change is a reshuffle you cannot tell from an improvement. …
Read the whole file at its exact version.
How to install
mdr add nahid-sparktales/agent-dispatcher/retrieval-rag@git:20260919.a0d4f55mdr add nahid-sparktales/agent-dispatcher/retrieval-rag@sha256:0f9790a151c040f1Pin to a label to follow the author's releases, or to a sha256 to freeze the exact bytes forever. Either way the resolved hash is written to mdr.lock, and mdr install reproduces it on any machine.
[](https://markdownregistry.com/a/art_3bwyh3ha3kssqwis)
1 badge views in 30 days
Versions
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (7096 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
nahid-sparktales/agent-dispatcher · 49 stars · license MIT · pushed 2026-09-23 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_3bwyh3ha3kssqwis GET https://markdownregistry.com/api/v1/resolve?ref=nahid-sparktales/agent-dispatcher/retrieval-rag GET https://markdownregistry.com/api/v1/blob/0f9790a151c040f1a93523c19996129e6ffc15625cdbd1670df7bc6447bb7f57
Your agent does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.