Added a worked example to the art article: eight works scored through bin/indicators, showing which indicators fired and how hard. Built bin/indicators (generic weighted-indicator scorer).
Added 'Is it art? Indicators, not definitions': a high-leverage indicator set for telling art from non-art, built on Gaut's cluster theory and weighted by diagnosticity, applying the research method to a question outside intelligence.
Added a research catalog (catalog.html): every article, the collected datasets with source grades (a snapshot from bin/catalog, reproduced live), and the tools. Also built bin/catalog to inventory all saved collections.
Published AGENTS.md for the research system (apps/intel/AGENTS.md, symlinked): how an agent should use the platform — the collect/grade/trace/challenge/ACH/forecast workflow and the rules, building on ruby.im's AGENTS.md. Linked from the hub.
Updated the tradecraft page: the analytic half of the platform (source grading, key-assumptions/premortem/red-team, an ACH engine, calibrated forecast scoring) is now built as tools, not just designed.
Applied analytic tradecraft back to the research: added 'Reanalysis: how this could be wrong' to the leverage map (visibility assumption; agent-native gold-rush competing hypothesis), a mirror-imaging/undercount note to the teams page, and a self-report source-reliability caveat to the builder profiles.
Added 'Intelligence tradecraft, amplified by AI': a deep dive on structured analytic techniques (ACH, key-assumptions check, indicators, premortem, calibrated forecasting) and ways of viewing the world, with what AI does to each. Linked from the hub.
Rebuilt the research.ruby.im front page as a hub that links to every page with a short description and a suggested reading path; moved the intel-stack writeup to /stack.html and repointed cross-links.
Added 'Every leverage pattern, tier-ranked': 23 distinct high-leverage patterns ranked S/A/B/C by multiplier, permissionlessness and durability, building on Naval's permissionless-leverage framework. Linked across the series.
Added 'The high-leverage teams': about 100 engineering teams grouped by leverage pattern, with a refined 'lessons from running the research system' section (meta-findings from a day of toolkit runs). Linked from the overview and setups pages.
Added 'How high-leverage builders actually work': Simon Willison, antirez, Mitchell Hashimoto, Linear, 37signals and Anthropic's Claude Code team, their documented tools and setups compared. Linked from the home page.
Launched research.ruby.im with the first article: building a token-cheap open-source intelligence stack in Ruby (RAG, a collection toolkit over free structured sources, Katana and GPT Researcher, SearXNG, an honest access policy, and three case studies).
Provided as is, for information only, with no warranty and no liability. Not legal or professional advice. Security and radio material is for authorized, lawful use only. Full disclaimer.
Visits are logged on our own server: IP address, browser and device details, and pages viewed. We use them to understand traffic and spot attacks, and never share or sell them.