| name | vault-connect |
| description | Bridge unrelated vault domains. Find structural analogues between different areas. The vault's link graph knows connections you haven't seen.
|
/vault-connect [topic A] [topic B]
Find hidden connections between seemingly unrelated vault domains.
Modes
- Two topics given: Find structural bridges between A and B.
- One topic given: Find the most surprising connections TO this topic from other domains.
- No topic given: Scan recent activity and find the highest-value cross-domain bridge.
Steps
-
Read _CLAUDE.md at vault root for folder map.
-
Read index.md for vault topology.
-
Map the domains -- for each topic:
- Search vault for all notes mentioning it (title, body, tags, links)
- Identify the domain cluster (which folders, which projects, which time period)
- Extract structural properties: is it a process? a framework? a failure mode? a trade-off?
-
Find bridges -- spawn 3 parallel bridge agents:
- Structural agent: Same abstract shape in different domains.
- Causal agent: Does one domain's output feed another? Does a finding explain a failure?
- Temporal agent: Things that happened simultaneously or in sequence.
-
Present connections ranked by surprise and utility:
## Connections Found
### 1. [Connection Name] (surprise: high/medium, utility: high/medium)
**Domain A:** [concept] -- from [[note]]
**Domain B:** [concept] -- from [[note]]
**Bridge:** [what they share structurally]
**Implication:** [what you can DO with this connection]
-
If a connection is strong enough: propose a synthesis note that explicitly links the domains.
-
Append to log.md: ## [YYYY-MM-DD] connect | [topic(s)] -- [N] bridges found
Rules
- All output in the user's language (from
_CLAUDE.md Language section). Translate ALL template headings: "## 发现的联系", "领域A：", "桥梁：", "启示：" for Chinese users. Cross-language connections (e.g., a Chinese note bridging to an English note) are especially valuable -- flag them.
- Surprise is the metric -- "two related notes are related" is boring
- Structural analogy, not surface similarity -- shared words don't count. Shared SHAPES count.
- Be honest about stretch -- label genuine insights vs creative reaching
- Cite specific notes -- verifiable connections
- Propose actions -- a connection without an implication is trivia
