construct-bridge-detect

Detects cross-domain connection candidates between knowledge cards and writes them to a JSON log.

Updated Apr 21, 2026
One-click install
npx skills add https://github.com/mabstruct/construct --skill construct-bridge-detect-mabstruct
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: construct-bridge-detect
Source: https://github.com/mabstruct/construct/tree/main/CONSTRUCT-CLAUDE-impl/claude/skills/construct-bridge-detect
Command: npx skills add https://github.com/mabstruct/construct --skill construct-bridge-detect-mabstruct

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Knowledge spread across multiple domains often hides genuine structural parallels that manual review misses. This Skill systematically scans a knowledge workspace to surface cross-domain bridge candidates so curators can connect related ideas instead of leaving them siloed. ## Core Features & Use Cases - Three-Level Detection: Finds structural bridges in connections.json, category overlaps across domains, and content-level similarity via cross_domain_links in domains.yaml. - Candidate Assessment: Evaluates each candidate for genuine parallel, relation type, strength, and novelty before proposing it. - Machine-Readable Output: Persists the full candidate set to log/bridge-candidates.json as a regenerable derived artifact, then applies confirmed bridges to connections.json with event logging. - Use Case: During a curation cycle, run bridge detection to discover that a cosmology card on observer effects parallels a philosophy-of-mind card on self-model observer loops, then confirm and record the new connection. ## Quick Start Ask the agent to find connections across domains in the current workspace and review the proposed bridge candidates.

Frequently Asked Questions about construct-bridge-detect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find connections across knowledge domains?▼

Run bridge detection during a curation cycle or by asking for cross-domain patterns. The Skill scans connections.json for structural bridges, checks shared content_categories across domains, and reads card content for linked domains before proposing candidates.

What are the three levels of bridge detection?▼

Level 1 finds structural bridges where cards connect across domains in connections.json. Level 2 detects category overlap where the same content_categories appear in multiple domains. Level 3 reads card content for domains with cross_domain_links to assess genuine parallels.

Where are bridge detection results saved?▼

Candidates are written to log/bridge-candidates.json in the workspace as a derived, regenerable artifact. The file is written on every run, even with an empty candidates array, and confirmed bridges are stored canonically in connections.json.

Does bridge detection modify my knowledge graph automatically?▼

No. The Skill only proposes candidates and asks for confirmation before adding edges to connections.json. Confirmed bridges are logged as detect_bridge events, and an optional connection-type card can document the parallel.

What are the limitations of automated bridge detection?▼

Structural and category-based signals can produce superficial keyword coincidences, so each candidate requires content-level assessment for genuine parallel, strength, and novelty. Only cross-domain candidates are kept, and already-confirmed bridges are excluded.