construct-help

Inspects workspace state and suggests the next knowledge-management action with a command menu.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Users of a personal knowledge workspace often don't know what to do next — whether to set up a domain, run research, curate cards, or write output. This Skill removes that ambiguity by scanning the actual workspace state and proposing the single most valuable next action. ## Core Features & Use Cases - State-aware suggestions: Checks domain directories, cards, connections, research freshness, and pending curation items in a fixed priority order to pick one primary suggestion. - Command menu: Always presents a grouped list of available commands (init, research, curate, gaps, write, validate, and more) so users know what's possible. - Multi-domain awareness: Compares all domain workspaces and leads with the one needing the most attention. - Use Case: A user opens a session and says "what's next?" — the Skill detects 12 cards with no connections and suggests running a curation cycle, then routes the accepted choice to the right workflow. ## Quick Start Ask the assistant "what can I do next in my knowledge workspace?" to get a state-based suggestion and the full command menu.

Frequently Asked Questions about construct-help

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

FAQPage Schema
How do I know what to do next in a knowledge workspace?▼

Trigger the help skill by saying "help" or "what's next?" and it scans your workspace state — domains, cards, connections, and research freshness — then proposes one primary next action you can accept or decline.

How does the skill decide which suggestion to show first?▼

It checks conditions in a fixed priority order and stops at the first match: no workspace, empty domain, missing interview, no connections, stale research, pending curation items, then growth suggestions for healthy graphs.

Does it work with multiple domain workspaces?▼

Yes. It checks all domains, leads with the one needing the most attention (emptiest, stalest, or most pending items), and briefly mentions secondary suggestions for other domains.

What happens after I accept a suggestion?▼

Accepting a suggestion runs the corresponding skill directly, picking a menu command routes to that skill, and unrelated input falls back to normal task routing defined in AGENTS.md.

Does the help skill block me from doing other tasks?▼

No. The suggestion is an offer, not a gate. You can ignore it, pick any command from the menu, or phrase requests in natural language however you want.