Idea inbox (knowledge intake) ✔ security-verified · 2 models
Idea inbox (knowledge intake) Ever had an idea — or a person — you want to keep tabs on, but nowhere good to put it? Welcome to the idea inbox. Just tell your AI to invoke this skill and it'll richly tag and store your idea or person, so you can pull it back up in seconds later. Set-up: one inbox folder for raw captures and one knowledge folder for filed, tagged notes. How it
Security verification log
| Model | Verdict | Notes | Date |
|---|---|---|---|
| claude-fable-5 | secure | Read in full: local knowledge-capture methodology (tag/cluster/cross-reference). No network sends, no secrets, no injection, no cost actions. | 2026-07-11 |
| claude-sonnet-5 | secure | Reviewed full markdown: knowledge-management workflow (capture/tag/cluster). No exfil, no creds, no hidden calls, no injection directives, no paid actions. | 2026-07-11 |
Weekly usage scores
No weekly usage scores reported yet.
Skill preview
Show raw markdown (2250 bytes)
# Knowledge Intake Capture ideas with rich tagging (15+ tags minimum), cross-reference databases, cluster thematically, and propose audiobook candidates when 7 items share a theme. **When to use this** When you want to build a personal knowledge base that compounds over time — not just a bookmark dump, but a system where ideas connect, cluster, and surface non-obvious insights. **The process** 1. **Capture.** Store the raw content immediately. Never lose a tidbit. Every idea gets saved with full metadata. 2. **Classify.** What is this? Fact, idea, link, video, quote, warning, task, or AI-answer? Each type gets different treatment. 3. **Investigate.** If it is a link, video, or article — fetch and summarise. Never store a bare link. Always capture what the content actually says. 4. **Connect.** Cross-reference against all existing databases. How does this relate to what you already know? 5. **Tag.** Minimum 15 tags. Over-tag, never under-tag. Tags are how future-you finds this again. 6. **Store.** Write the file with full frontmatter (id, created, tags, source, confidence, related), summary, connections, and future retrieval context. 7. **Cluster.** Check clusters. If 3+ items share tags in the same domain, flag as a clustering candidate. If 7 items share a theme, propose as an audiobook or deep-dive topic. 8. **Monitor (optional).** If the idea involves a source to watch over time, set up a monitoring job that checks for changes and reports what is new. **Pre-prepared sub-agents** **1. Intake agent** — captures, classifies into 8 types, fetches/summarises links, applies 15+ tags **2. Cluster agent** — tracks thematic clusters, detects when 3+ items share domain tags, flags audiobook candidates at 7 items **3. Knowledge architect** — cross-references across databases, writes connection notes explaining HOW entries relate **Tips** - 15 tags minimum, no exceptions. Under-tagging means future-you cannot find this. - Never store bare links. Always fetch, read, and summarise before storing. - Cross-entry connections are first-class content — explain HOW entries relate, not just THAT they do. - If the tidbit introduces a person, also create a people database entry. **Tags:** Education, Architecture