prime

Load ADR and openspec documents into conversational context with semantic topic filtering.

30|2|Updated Feb 13, 2026

Search Tags:#architecture #adr #spec #context-loading #semantic-matching #design-governance

Ready to run

npx skills add https://github.com/joestump/claude-plugin-design --skill prime-joestump

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: prime
Source: https://github.com/joestump/claude-plugin-design/tree/main/skills/prime
Command: npx skills add https://github.com/joestump/claude-plugin-design --skill prime-joestump

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Load existing architecture decisions and formal specifications into Claude's conversational context so that design-aware guidance, checks, and summaries can be provided without repeatedly re-explaining project decisions.

Core Features & Use Cases

Context aggregation: Scans project ADRs and openspec specs, extracts YAML frontmatter, titles, context, outcomes, and key sections for quick consumption.
Semantic topic filtering: Allows targeted priming by matching a user-provided topic to relevant ADRs and specs using semantic similarity rather than simple keyword search.
Practical workflows: Use to prepare the assistant for design reviews, drift checks, documentation generation, or to start a new architecture-aware session that references past decisions.

Quick Start

Run /design:prime security to load ADRs and specs related to the topic "security" into the session.

Frequently Asked Questions about prime

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

FAQPage Schema

How do I load ADR and openspec specification documents into a session for architecture-aware responses?▼

To load ADR and openspec specification documents, the Skill scans repositories storing MADR ADRs in docs/adrs and openspec specs in docs/openspec/specs, parsing YAML frontmatter and headings to inject context for architecture-aware responses.

What is the best way to find relevant architecture decisions for a specific topic like security?▼

Finding relevant architecture decisions for a specific topic uses semantic similarity filtering rather than simple keyword search, matching your provided topic argument to relevant ADRs and specs for focused design reviews.

Can I use this to detect specification drift across my project documentation?▼

Yes, you can detect specification drift by loading existing ADRs and openspec specs into the session, enabling the assistant to compare past decisions against current code or documentation for inconsistencies.

Does this require a specific directory structure to parse markdown artifacts?▼

It requires read-only repository access and expects MADR ADRs stored in docs/adrs and openspec specs in docs/openspec/specs to reliably extract YAML frontmatter, titles, and specified sections from markdown artifacts.

How do I prepare an assistant for a design review using existing project decisions?▼

Prepare for a design review by running a prime command with a topic argument, which aggregates context and outcomes from existing ADRs and specs so the assistant references past decisions without re-explanation.

What are the limitations of using semantic matching for loading architecture context?▼

Limitations include dependency on reliable YAML frontmatter parsing and heading extraction from markdown artifacts, meaning poorly formatted or non-standard ADR and spec files may not load correctly into the conversational context.