write-ai-memex - SKILL.md Agent Skill

name: write-ai-memex description: "Write knowledge entries to the AI Memex — capture patterns, research, decisions, and project outcomes during any coding session"

Write AI Memex Entry

You are helping the user capture knowledge in their AI Memex — a distributed knowledge system published at lqdev.me/resources/ai-memex/.

Context Detection

Before writing, detect where you are:

Check for PersonalSite.fsproj in the current directory or any parent directory
If found → you're in the lqdev.me repo → write directly to _src/resources/ai-memex/{slug}.md
If not found → you're in another project → write to .ai-memex/{slug}.md in the project root

Entry Types

Type	Purpose	Use When
`pattern`	Reusable solutions, gotchas, proven approaches	Bug fix, architecture decision, recurring solution
`research`	Technology evaluations, comparisons	Multi-approach analysis completed
`reference`	Living docs, architecture overviews	Something reusable was built
`project-report`	Feature summaries, retrospectives	Feature shipped, milestone reached
`blog-post`	AI-human collaboration insights	Meta-observations about the process

File Naming

Use kebab-case slugs that describe the content:

pattern-progressive-loading.md
research-caching-strategies.md
reference-api-auth-patterns.md
project-report-v2-migration.md

YAML Frontmatter Schema

---
title: "Clear, Descriptive Title"
description: "One-sentence summary of what this entry covers"
entry_type: pattern          # One of: pattern, research, reference, project-report, blog-post
published_date: "YYYY-MM-DD HH:mm zzz"
last_updated_date: "YYYY-MM-DD HH:mm zzz"
tags: "tag1, tag2, tag3"     # Comma-separated, lowercase
related_skill: ""            # Optional: skill name if this pattern relates to a skill
source_project: ""           # Optional: project name where this was discovered
---

Quality Standards

All entries

Evidence-based: Include specific code, commands, or examples
Context-rich: Explain WHY, not just WHAT
Discoverable: Use clear titles and relevant tags
Self-contained: Readable without other context

By type

pattern: Include Discovery (what happened), Root Cause (why), Solution (how), and Prevention (avoid in future)
research: Include Options Considered, Evaluation Criteria, Recommendation, and Trade-offs
reference: Include Overview, Key Components, Usage Examples, and Gotchas
project-report: Include Objective, Approach, Outcome, and Lessons Learned
blog-post: Include Observation, Context, Reflection, and Takeaway

Consent Protocol

ALWAYS ask before creating. Never auto-generate entries.

Propose like this:

"This looks like a good [type] entry for the AI Memex. Want me to write it up?"

If the user says yes, write the full entry. If they want changes, iterate.

Tag Taxonomy

Use tags from these categories:

Technology: fsharp, dotnet, python, typescript, javascript, rust, docker, kubernetes, azure
Domain: web, api, databases, devops, architecture, security, performance, accessibility
Meta: patterns, research, ai-collaboration

See references/TEMPLATES.md for per-type frontmatter templates.