How do I install the Haiku RAG skill?

Run npx skillmds@latest add ggozad/haiku-rag in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Haiku RAG safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Haiku RAG?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Haiku RAG free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Haiku RAG?

ggozad (@ggozad) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Haiku RAG by ggozad — Agent Skill

Name: Haiku RAG
Author: ggozad

Haiku RAG

Search, read and compute over the user's haiku.rag knowledge base through the haiku-rag MCP tools. Use whenever a request could be answered from the user's ingested documents, when asked to find, look up, check or cite something in their documents or knowledge base, or when the question is about the user's own material rather than general knowledge.

ggozad 532186d 3.2 KB Updated Sep 17, 2026

File contents

Working with the knowledge base

Check the knowledge base before answering from memory whenever the question could be about the user's documents. Say so when it has nothing relevant.

Find

search_documents is the first call. Results come best first with the document title, section headings, the matched chunk's metadata when it has any, and the passage in its section. Pictures in the results arrive as images: answer figure questions from them. filter restricts which documents are searched, limit how many results come back. If it misses, rephrase once or narrow with a filter before concluding the material is not there. When the question is about an image rather than words and the server offers search_documents_by_image, it takes the image as the query.

Read

Every search result shows its Document ID (and Collection when there are several); pass them to the read tools. get_document returns a document's whole text in reading order. For a long one, get_document_outline gives the heading tree with page numbers and get_document_section the text of one section, subsections included.

Compute

execute_code runs a Python program on the server over the same documents. Under /documents/{id}/ each has metadata.json, content.txt, items.jsonl, chunks.jsonl and toc.json, and the program can await search(query) and await list_documents(). Write code when the answer is a count, an aggregate, a comparison across many documents, a lookup by document or chunk metadata, or a pattern over whole documents: whatever search cannot rank. Each call is one program and variables do not carry over, so gather, compute and print a compact result in the same program. filter and sources select the documents it sees. For a known document's structure read its toc.json first; search() ranks across every document. Map a title or URI to an id with one list_documents() call rather than reading every metadata.json; the files carry no source, so over several collections group by its rows. Answer and cite from what it printed.

Explore

list_documents shows what is stored: titles, URIs and metadata. It is how you learn what a filter can match.

Filters

A SQL WHERE clause over the document columns id, uri, title, created_at, updated_at, metadata. metadata is a JSON string, so match it with LIKE: metadata LIKE '%"author": "Smith"%'. Also uri LIKE '%.pdf', title = 'Q3 report'.

Results and citations

Rank is the signal; scores are not comparable across queries and are never confidence. Cite the document title or URI, the section heading and page numbers when present, and the matched chunk's metadata when it carries locators such as paragraph or footnote numbers. When results carry source, the server covers several collections: name it, and pass sources to search a subset.

---name: haiku-ragdescription: Search, read and compute over the user's haiku.rag knowledge base through the haiku-rag MCP tools. Use whenever a request could be answered from the user's ingested documents, when asked to find, look up, check or cite something in their documents or knowledge base, or when the question is about the user's own material rather than general knowledge.---# Working with the knowledge baseCheck the knowledge base before answering from memory whenever the questioncould be about the user's documents. Say so when it has nothing relevant.## Find`search_documents` is the first call. Results come best first with the documenttitle, section headings, the matched chunk's metadata when it has any, and thepassage in its section. Pictures in the results arrive as images: answerfigure questions from them. `filter` restricts which documents are searched,`limit` how many results come back. If it misses, rephrase once or narrow witha filter before concluding the material is not there. When the question isabout an image rather than words and the server offers`search_documents_by_image`, it takes the image as the query.## ReadEvery search result shows its `Document ID` (and `Collection` when there areseveral); pass them to the read tools. `get_document` returns a document'swhole text in reading order. For a long one, `get_document_outline` gives theheading tree with page numbers and `get_document_section` the text of onesection, subsections included.## Compute`execute_code` runs a Python program on the server over the same documents.Under `/documents/{id}/` each has `metadata.json`, `content.txt`, `items.jsonl`,`chunks.jsonl` and `toc.json`, and the program can `await search(query)` and`await list_documents()`. Write code when the answer is a count, an aggregate, acomparison across many documents, a lookup by document or chunk metadata, or apattern over whole documents: whatever search cannot rank. Each call is oneprogram and variables do not carry over, so gather, compute and `print` acompact result in the same program. `filter` and `sources` select the documentsit sees. For a known document's structure read its `toc.json` first; `search()`ranks across every document. Map a title or URI to an id with one`list_documents()` call rather than reading every `metadata.json`; the filescarry no `source`, so over several collections group by its rows. Answer andcite from what it printed.## Explore`list_documents` shows what is stored: titles, URIs and metadata. It is how youlearn what a filter can match.## FiltersA SQL WHERE clause over the document columns `id`, `uri`, `title`,`created_at`, `updated_at`, `metadata`. `metadata` is a JSON string, so matchit with LIKE: `metadata LIKE '%"author": "Smith"%'`. Also `uri LIKE '%.pdf'`,`title = 'Q3 report'`.## Results and citationsRank is the signal; scores are not comparable across queries and are neverconfidence. Cite the document title or URI, the section heading and pagenumbers when present, and the matched chunk's metadata when it carries locatorssuch as paragraph or footnote numbers. When results carry `source`, the servercovers several collections: name it, and pass `sources` to search a subset.

Frequently asked questions