Agent skill

Give this teaching to a code-writing machine before it makes asking-screens. It helps the machine choose the right pattern and avoid dangerous yes-boxes.

The Agent Consent Patterns skill gives a coding agent the same vocabulary, selection guidance, review checklist, and accessibility rules used throughout this site. Use it when an agent is building or reviewing permissions, approvals, standing authority, credential handoff, or action receipts.

The Agent Consent Patterns skill operationalizes this reference for coding agents: it supplies pattern-selection criteria, non-negotiable consent constraints, anti-patterns, and accessibility checks for delegated-authority interfaces.

After it is installed, ask the agent to use it, or just ask it to build an asking-screen. The skill can wake up when it fits.

After installation, invoke the skill explicitly when needed or let the agent select it when a task matches its description. Use it before implementation, during design review, or when auditing an existing consent flow.

Installed skills may be invoked explicitly or selected from their trigger descriptions. Apply this one at design time, implementation time, and review time so the consent contract remains consistent across the workflow rather than becoming a final-pass checklist.

Copy the whole skill folder, not just one file, into the place your agent looks for skills.

For another agent that supports the open SKILL.md format, copy the complete plugins/agent-consent-patterns/skills/agent-consent-patterns/ directory into that agent's documented skills location. Preserve the directory name and front matter; installation locations differ by product.

The skill is distributed in the interoperable SKILL.md directory format. Compatible agents can consume the complete plugins/agent-consent-patterns/skills/agent-consent-patterns/ directory through their own documented discovery mechanism; platform-specific installation paths are intentionally not assumed to be interchangeable.