kb

Search local knowledge bases for lessons and API gotchas.

1|Updated Feb 15, 2026
One-click install
npx skills add https://github.com/n0rvyn/indie-toolkit --skill kb-n0rvyn
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: kb
Source: https://github.com/n0rvyn/indie-toolkit/tree/main/dev-workflow/skills/kb
Command: npx skills add https://github.com/n0rvyn/indie-toolkit --skill kb-n0rvyn

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Quickly locate past lessons, API gotchas, architecture decisions, and learning notes across a user's local cross-project knowledge base to avoid repeating past mistakes and to accelerate troubleshooting.

Core Features & Use Cases

  • Search both content and frontmatter keywords across the centralized ~/.claude/knowledge/ tree with optional category scoping.
  • Classify queries as question-mode or browse-mode, show grouped file results with freshness indicators, and offer to read full entries.
  • For question-style queries, read top matches and synthesize a concise answer citing source filenames; fall back to project-local docs and an Obsidian PKOS vault when needed.
  • Use case: a developer investigating a recurring SwiftData concurrency bug can run a single query to find prior postmortems, mitigations, and architecture notes.

Quick Start

Use the kb skill by entering: /kb How do I handle SwiftData concurrency across actors?

Frequently Asked Questions about kb

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search my local knowledge base for past engineering lessons and API gotchas?

Search cross-project knowledge by running content and keyword Grep queries across the ~/.claude/knowledge/ tree to instantly locate past engineering lessons, API gotchas, and architecture decisions. The Skill classifies your query intent and groups file results with freshness indicators.

What's the best way to find previous architecture decisions across multiple project docs?

Find previous architecture decisions across multiple project docs by applying configurable category scoping and freshness ranking by frontmatter dates to your cross-project knowledge base search. It reads top matches to synthesize a concise answer citing source filenames.

Can I use Obsidian PKOS vault lookups when searching cross-project knowledge?

Obsidian PKOS vault lookups are supported as an optional fallback when searching cross-project knowledge. When handling question-style queries, the Skill falls back to the ~/Obsidian/PKOS/ directory and project-local docs if initial knowledge base results are insufficient.

How does freshness ranking by frontmatter dates work for local docs search?

Freshness ranking by frontmatter dates works by sorting local docs search results based on date metadata stored in file frontmatter. This ensures the most recently updated knowledge base entries appear first, helping you find current mitigations and postmortems instead of outdated notes.

Does the cross-project knowledge search work with project-local docs/09-lessons-learned?

Cross-project knowledge search works with project-local docs/09-lessons-learned as an optional fallback. When the centralized ~/.claude/knowledge/ tree does not yield sufficient results for question-style queries, the Skill automatically checks this project-local directory for relevant entries.

When should I not rely on a local knowledge base search for troubleshooting?

Do not rely on local knowledge base search when your ~/.claude/knowledge/ repository is empty or lacks frontmatter dates for freshness ranking. Search effectiveness depends entirely on existing local notes, meaning completely undocumented issues will return no synthesized answers or file matches.