connector-hub

Coordinates read-only connectors like Jira, GitHub, DB, Notion via a JSON envelope of plans and results.

10|Updated Apr 28, 2026
One-click install
npx skills add https://github.com/narailabs/narai-primitives --skill connector-hub
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: connector-hub
Source: https://github.com/narailabs/narai-primitives/tree/main/plugin/skills/connector-hub
Command: npx skills add https://github.com/narailabs/narai-primitives --skill connector-hub

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Connector Hub helps you answer a question that spans multiple read-only connectors by turning one natural-language prompt into an organized set of connector actions executed concurrently.

Core Features & Use Cases

  • Multi-connector context gathering: Plans and dispatches work across Jira, GitHub, DB, Notion, and other read-only connectors from a single prompt.
  • Parallel execution: Runs each planned step in parallel to reduce time-to-answer.
  • Structured planning output: Returns a JSON envelope containing both the plan and per-step results, including inline errors.

Quick Start

Ask the connector-hub binary to fulfill your request by providing your question in a single prompt, then read the returned JSON envelope to access the plan and results.

Frequently Asked Questions about connector-hub

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

FAQPage Schema
How do I gather context across Jira and GitHub from a single natural-language prompt?

Multi-connector context gathering from a single natural-language prompt is orchestrated by planning and dispatching read-only calls across systems like Jira and GitHub. This process runs each planned step in parallel to produce a JSON envelope containing the plan and per-step results.

Can I run read-only research queries in parallel across multiple data sources?

Yes, parallel execution of read-only research queries is supported. The system plans multiple connector actions concurrently across data sources, reducing time-to-answer by dispatching all steps simultaneously and returning a structured JSON envelope with inline errors.

What's the best way to plan multi-system research across Notion and a database?

Planning multi-system research across Notion and a database is best handled by submitting one natural-language prompt that the orchestrator uses to build a structured plan. It dispatches read-only connector calls in parallel, returning a JSON envelope with the plan and per-step results.

Does connector-hub enforce read-only guardrails when querying Jira or GitHub?

Read-only guardrails are strictly enforced during all connector calls. The orchestrator ensures that actions planned and dispatched across systems like Jira or GitHub remain read-only, inlining any errors directly into the results array of the returned JSON envelope.

How do I handle inline errors when dispatching parallel connector calls?

Inline errors from parallel connector calls are captured directly within the results array of the JSON envelope. The structured planning output includes both the successful per-step results and the inline errors, allowing you to review the complete execution plan and outcomes.