A searchable index of ~440 posts from John Cutler's The Beautiful Mess newsletter, packaged as an open SKILL.md for AI coding agents.
When you ask your AI agent about product management, org design, strategy, prioritization, metrics, delivery, leadership, or systems thinking, this skill searches the TBM archive and returns the most relevant posts with links.
Clone this repo into your skills directory:
git clone https://github.com/johnpcutler/tbm-recommender.git ~/.cursor/skills/tbm-recommenderThe skill activates automatically when you mention TBM, John Cutler, or related topics.
git clone https://github.com/johnpcutler/tbm-recommender.git ~/.claude/skills/tbm-recommenderAny agent that supports the SKILL.md format can use this skill. Clone the repo into your agent's skills directory.
The skill uses a three-tier retrieval flow to keep token usage low:
- Route -- The agent reads a compact cluster index (~600 tokens) to identify which topic areas match your query.
- Scan -- It loads only the matching cluster's posts and compares your intent against pre-extracted questions and search terms.
- Explore -- Optionally follows a similarity graph to find related posts.
SKILL.md # Agent instructions (open SKILL.md format)
data/
clusters.json # 21 topic clusters (routing layer)
clusters/*.jsonl # Per-cluster post details (questions, search terms, summary)
content-graph.json # Post-to-post similarity neighbors
version.json # Build metadata
The tiered design keeps costs low. Approximate input tokens per query:
| Scenario | Tokens |
|---|---|
| Tier 1 only (routing) | ~1,600 |
| Tier 1 + 1 small cluster | ~3,300 |
| Tier 1 + 1 average cluster | ~8,800 |
| Tier 1 + 2 large clusters | ~29,000 |
A typical query hits 1-2 clusters and costs 3,000-15,000 input tokens. The similarity graph (Tier 3) is only loaded when the user asks for related posts.
Product management, discovery, experimentation, roadmaps, planning, prioritization, strategy, OKRs, metrics, WIP limits, dependencies, delivery, org design, operating models, change management, continuous improvement, leadership, psychological safety, communication, trust, frameworks, systems thinking, AI at work, and more.
MIT