← Back to Skills Marketplace
Personal Knowledge Base Lite
by
congshengwu
· GitHub ↗
· v1.0.0
· MIT-0
204
Downloads
1
Stars
0
Active Installs
4
Versions
Install in OpenClaw
/install personal-kb-lite
Description
Local file knowledge base with LLM-powered indexing and Q&A. Scans a directory for documents (.txt, .md, .pdf, .docx, .xlsx, .csv), extracts metadata (summar...
Usage Guidance
This skill appears to do what it says: build a local index and answer questions from files. Before installing:
- Do not point WATCH_DIR to system folders, your home directory, or any location containing secrets; create a dedicated directory with only the documents you want indexed.
- Confirm what the installer 'uv' actually does (ask the publisher or registry) and prefer installing known packages via pip from PyPI or in a virtual environment. Review package sources/version and hashes if possible.
- Run the skill in a sandboxed environment or VM first so indexing/writes can't affect important files.
- Be aware the agent will read file contents to generate summaries — treat that as sensitive access and restrict who can invoke the skill. If you need stricter guarantees (no accidental data reads), require manual invocation and review the config before running.
- If the registry/homepage/author is unknown, ask the publisher for provenance or source code before granting any automated install rights.
Capability Analysis
Type: OpenClaw Skill
Name: personal-kb-lite
Version: 1.0.0
The skill implements a local document indexing and RAG (Retrieval-Augmented Generation) system. It scans a user-defined directory for documents, extracts metadata using the LLM, and stores it in a local JSON file. The implementation uses safe practices, such as passing file paths via environment variables to Python sub-processes to prevent shell injection, and its behavior is entirely consistent with its stated purpose of local knowledge management.
Capability Assessment
Purpose & Capability
The name/description (local KB, indexing PDFs/DOCX/XLSX/CSV/md/txt) aligns with required binaries (python3) and the listed Python packages (pdfplumber, python-docx, openpyxl) which are appropriate for parsing those formats. No unrelated credentials, binaries, or config paths are requested.
Instruction Scope
Instructions explicitly read/write .kb-config.json and .kb-meta.json and scan a user-provided WATCH_DIR using find/stat/wc; this is appropriate for indexing. However, because the tool asks the user for an absolute path and will operate on any path provided, the user must avoid pointing it at sensitive system or home directories. There are no instructions to transmit data externally, but the agent’s Read/Write tools and LLM summarization will access file contents.
Install Mechanism
The install entries list pdfplumber, python-docx, and openpyxl — reasonable for the stated task and typically obtained from PyPI. The install kind is labeled 'uv', which is nonstandard/ambiguous in this manifest (not a well-known installer name like 'pip' or a direct GitHub release); clarify the install mechanism before installing so you know what will be fetched and executed.
Credentials
The skill requests no environment variables, credentials, or external tokens. It only reads/writes local config and metadata files in the working directory and WATCH_DIR, which is proportionate to a local KB.
Persistence & Privilege
The skill is not marked always:true and does not request system-wide changes or other skills' configurations. It writes its own .kb-config.json/.kb-meta.json only. Autonomous invocation is allowed by default (normal for skills) but not otherwise elevated.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install personal-kb-lite - After installation, invoke the skill by name or use
/personal-kb-lite - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
personal-kb-lite 1.0.0 – Initial release
- Create and update a local knowledge base by indexing documents in a chosen directory.
- Supports file types: .txt, .md, .pdf, .docx, .xlsx, .csv.
- Uses LLM to extract per-file summaries, tags, and table of contents for rapid search.
- Answers user questions by retrieving and synthesizing content from relevant files, with citations.
- Includes metadata management and progress tracking for file indexing.
v0.0.3
- Adds support for custom command aliases via --alias and --alias reset.
- Recursively indexes files in all subdirectories, not just top-level.
- Detects and re-indexes modified files; cleans up deleted/stale metadata entries.
- Uses relative paths for unique file keys instead of just filenames.
- Simplifies configuration and command (default is now /kb), and updates all messages accordingly.
- Improves reliability and resiliency of indexing (immediate saves, detects changes, error handling).
- Updates documentation and metadata for clarity.
v0.0.2
- Renamed the skill from "local-kb" to "personal-kb-lite".
- Changed the command and usage instructions to `/personal-kb-lite` for both indexing and querying.
- Updated user-facing prompts and error messages to reflect the new command name.
- No changes to core workflows or functionality.
v0.0.1
personal-kb-lite 1.0.0
- Initial release of local-kb: a local file knowledge base with LLM-powered indexing and question answering.
- Supports indexing of .txt, .md, .pdf, .docx, .xlsx, and .csv formats.
- Automatically extracts summaries, tags, and table of contents for each document.
- Provides easy setup of watch directory and robust progress reporting during indexing.
- Users can query the knowledge base for answers, with relevant files identified and analyzed.
Metadata
Frequently Asked Questions
What is Personal Knowledge Base Lite?
Local file knowledge base with LLM-powered indexing and Q&A. Scans a directory for documents (.txt, .md, .pdf, .docx, .xlsx, .csv), extracts metadata (summar... It is an AI Agent Skill for Claude Code / OpenClaw, with 204 downloads so far.
How do I install Personal Knowledge Base Lite?
Run "/install personal-kb-lite" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Personal Knowledge Base Lite free?
Yes, Personal Knowledge Base Lite is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Personal Knowledge Base Lite support?
Personal Knowledge Base Lite is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Personal Knowledge Base Lite?
It is built and maintained by congshengwu (@congshengwu); the current version is v1.0.0.
More Skills