pi-llm-wiki

Overview

The pi-llm-wiki package acts as an intelligent intermediary layer for AI coding assistants like pi and oh-my-pi, transforming raw inputs such as URLs, documents, and code snippets into a durable, interconnected wiki. By utilizing immutable source packets, automated ingestion, layered vault architecture, and Open Knowledge Format (OKF v0.2) support, it enables persistent knowledge retention across projects and personal sessions. It also features opt-in agent working-memory tracking, deterministic linter health checks, and dual-host compatibility through a robust suite of tools and slash commands.

Capabilities

• ▸Dual-host support for pi and oh-my-pi coding agents
• ▸Immutable source capture for URLs, PDFs, markdown, JSON, and XML
• ▸Automated ingestion pipeline into concept, entity, synthesis, and analysis pages
• ▸Layered vault architecture supporting personal and project-level knowledge bases
• ▸Obsidian-compatible folder-qualified wikilinks and Open Knowledge Format v0.2 support
• ▸Agent trajectory capture and skill distillation for reusable working memory
• ▸Deterministic linter and automated metadata rebuilding

Best for

Building and maintaining a persistent, compounding project knowledge base or personal documentation vault, Ingesting research papers, documentation, URLs, and local files into structured markdown knowledge pages, Querying past project decisions, architectural patterns, and reusable agent skills across sessions, Automating documentation maintenance, health checks, and linting for missing links or orphan pages

Works with

Security considerations

The skill includes internal guardrails that block direct editing of raw source files and generated metadata via tool calls, ensuring integrity of the knowledge base.

Categories