Practical overview
When it fits
Use it when a growing collection of documents or decisions should become reusable context for AI agents, rather than being reintroduced in each session.
Who it is for
- People using supported coding agents who want project context or research to persist across sessions.
- Researchers and teams collecting articles, papers, meeting notes, transcripts or project decisions over time.
- Developers who want Markdown-based knowledge files that remain readable and editable across agent tools.
Documented uses
- Convert source documents and notes into linked knowledge pages.
- Search accumulated material and answer questions with citations to wiki pages.
- Preserve useful answers and project context for future agent sessions.
- Maintain links, summaries and existing pages as new sources are added.
Requirements & limits
The project is open source under the MIT license. The documented tool is installed through agent plugin or skill workflows and requires uv before wiki initialization.
- Wiki scope is selected at initialization: global across projects or project-specific.
- The default semantic search is local; initialization installs FastEmbed, sqlite-vec, PyYAML and a local embedding model.
- Claude Code receives the full plugin path, while other listed agents use the skill; some compatible runtimes require manual copying or symlinking.
- Wiki initialization and runtime setup are separate from installing the plugin or skill.
- The project documentation recommends a regular database when the primary need is structured records and transactions.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Initialize a personal or project wiki
Set up the documented wiki structure and local runtime before adding source material.
Before you start
- Install the LLM Wiki plugin or skill for a supported agent.
- Install uv before running initialization.
Steps
Run
/wiki:init --globalfor a personal wiki shared across projects, or/wiki:initfor a wiki in the current project.Review the proposed agent-memory location pointer; the skill requires approval before writing to a memory file.
Check the initialization JSON runtime report for
"status": "ready".
Expected outcome
A selected wiki root is bootstrapped with source storage, schema, index, log and page template, and its local runtime is set up.
Guide limitations
- This covers setup only; ingesting and querying sources are separate workflow stages.
- Initialization does not write the proposed agent-memory pointer without approval.
Ingest a source and query the wiki
Add a source to an initialized wiki and ask a later question against its accumulated pages.
Before you start
- An initialized wiki with a ready runtime.
- A source file placed in the configured raw-source root.
Steps
Run
/wiki:ingestwith the source file path, such as/wiki:ingest ~/wiki/raw/your-source.pdf.Review the agent's source takeaways and proposed summary, entity and concept page updates.
Once sources are present, run
/wiki:queryfollowed by a question about the wiki's contents.
Expected outcome
The agent organizes the source into linked Markdown pages and can answer a query with wikilink citations; it may offer to save substantive answers in the wiki.
Guide limitations
- The documented example describes an agent reviewing takeaways with the user; outcomes depend on the source and interaction.
- The query stage requires source material to have been added first.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 13 October 2026
- Original website title
- praneybehl/llm-wiki-plugin · Official homepage title · retained in its source language