← ClaudeAtlas

intelligence-enginelisted

Run one weekly cycle of the intelligence engine — grade open predictions, research the pillars, score signals, rebuild the dashboard and write what was learned back to the knowledge base. Use when asked to "run the engine", "run cycle N", or refresh the intelligence dashboard.
ruhisalhotrahq/intelligence-engine · ★ 0 · AI & Automation · score 72
Install: claude install-skill ruhisalhotrahq/intelligence-engine
# Intelligence Engine You are running one cycle of a weekly intelligence engine. You have no memory of previous cycles except what is written in this folder. Read this file first, then follow it exactly. ## The question this engine answers > [Write one narrow question a weekly answer can serve. Example: "Which companies in our market are about to need what we sell, and what is changing in how our buyers decide?"] **Who reads the output:** [e.g. founders and the sales team; leadership; just me] **What the output feeds:** [e.g. sales outreach, published content, a weekly decision memo] ## Pillars The standing sections of every output and the lanes research runs in. Keep to three or four. 1. **[Pillar 1]** — [one line on what belongs here] 2. **[Pillar 2]** — [one line] 3. **[Pillar 3]** — [one line] 4. **[Pillar 4]** — [one line] ## Cadence Weekly, [day] at [time, timezone]. Before anything else, **check the system clock** and state the real date. Never trust a date written in your context or in these files. ## The cycle, in order The order is not optional. Grading before reading new research is what keeps the engine honest. 1. **Review every open prediction** — before any new research. Follow `KNOWLEDGE_GRAPH.md` → *Prediction review*. 2. **Mine the internal corpus** — [list your own inputs: call notes, reply logs, competitor watches, CRM exports]. State plainly which inputs contributed nothing this cycle, and which predictions that blocks. 3. **Run live research**