Skip to content
Get packs

Local Knowledge Base Retriever

Searches a local folder of docs, PDFs and Excel files to answer questions with source citations

Source Download skill
Repository
ConardLi/garden-skills
Stars
13,484
Licence
MIT
Files
8 files

What it does

Once installed, the agent treats a local folder (by default knowledge/, or a path you give it) as a knowledge base. It reads the folder's data_structure.md index files to find the most relevant subfolders and files, then searches them piece by piece with grep and partial reads instead of loading whole files. For PDFs and Excel workbooks it first converts or reads them with tools such as pdftotext, pdfplumber and pandas, then answers with the file names and approximate locations it used.

Great for

  • Teams that keep internal documents, reports or specs in a structured local folder and want quick, sourced answers
  • Analysts who need specific figures from large Excel workbooks without loading the whole file
  • People who want the agent to stop and ask for the right folder rather than guessing

Skip it if

  • Searching the open web or the internet: the skill tells the agent not to use web search, so it only covers local files (answers are limited to what is on disk)
  • Knowledge folders without data_structure.md index files: the search relies on those indexes to choose files, so results may be weaker or slower without them
  • Confidential data sent to an external AI service: the agent reads file contents into its conversation, so sensitive files should be reviewed before use

Try asking

  • “From the knowledge base, find the API timeout settings and tell me which document they are in”
  • “Look in the Q1 2023 sales report and give me the total revenue by region”
  • “Summarise the architecture decisions described in our design docs folder”
  • “Search the ./docs folder for anything about the refund policy and cite the sources”
Step-by-step guide: how to use Local Knowledge Base RetrieverInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 8 original files, the MIT licence and an attribution note.

    kb-retriever.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip kb-retriever.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip kb-retriever.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Start with a clear index: add data_structure.md files describing each folder so the agent can pick the right files quickly
  • Name the folder or file in your question (for example ./docs or the workbook name) to skip the search for the right location
  • Check the cited file names and locations in the answer, since the search is capped at 5 rounds and can miss content

What's inside

  • manifest.json
  • README.md
  • README.zh-CN.md
  • references/excel_analysis.md
  • references/excel_reading.md
  • references/pdf_reading.md
  • scripts/convert_pdf_to_images.py
  • SKILL.md

Redistributed unmodified from ConardLi/garden-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.