responsibility-gate

Defines a per-invocation user consent gate for skills that install or modify software.

Updated Mar 30, 2026
One-click install
npx skills add https://github.com/rubrical-works/idpf-praxis-skills --skill responsibility-gate-rubrical-works
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: responsibility-gate
Source: https://github.com/rubrical-works/idpf-praxis-skills/tree/main/Skills/responsibility-gate
Command: npx skills add https://github.com/rubrical-works/idpf-praxis-skills --skill responsibility-gate-rubrical-works

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Skills that install packages, edit PATH, or change system state can act on a user's machine without explicit consent. This Skill provides a single, authoritative contract requiring every install-capable or modify-capable skill to present a responsibility-acknowledgement gate before executing any system-changing action. ## Core Features & Use Cases - Gate Contract: Specifies when the gate fires, the exact AskUserQuestion wording, decline behavior, and the per-invocation (never persisted) acknowledgement rule. - Compliance Audit List: Enumerates in-scope skills (install-node, flask-setup, playwright-setup, and others) plus borderline reclassification decisions. - Authoring Guidance: Explains how to reference the gate from another SKILL.md without duplicating the contract, keeping a single source of truth. - Use Case: When authoring a new skill that runs npm install or edits shell rc files, link to this contract and add a responsibility-acknowledgement step so decline exits cleanly with zero system changes. ## Quick Start Read the responsibility-gate skill contract and add a responsibility acknowledgement step to my new install-capable skill before its first execution step.

Frequently Asked Questions about responsibility-gate

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

FAQPage Schema
How do I add a user consent prompt before a skill installs software?▼

Add a responsibility-acknowledgement gate step before the first execution step that uses AskUserQuestion with accept and decline options. The accept option must require explicit selection, and decline must exit cleanly with zero system changes.

Which skills must implement the responsibility-acknowledgement gate?▼

Any skill that installs software, modifies system state such as PATH or shell rc files, uninstalls previously installed software, or applies non-reversible configuration changes. Read-only operations like status checks and version probes must not trigger the gate.

Can a skill remember user consent between invocations?▼

No. Acknowledgement must be per-invocation and never persisted to disk, environment, session storage, or cache. Every subsequent invocation proposing an execution path must present the gate again so consent is never stale.

What happens when AskUserQuestion is unavailable in a sandboxed environment?▼

The skill must not execute the gated action. It should report the recommended command and the fact that the gate could not fire, then exit cleanly rather than proceeding without consent.

Should the responsibility gate be combined with command confirmation prompts?▼

No. The responsibility gate is independent of any command-confirmation step. A skill needing both must present two separate prompts, responsibility first, because bundling them obscures what the user is agreeing to.