Premium skill

Write prompt contracts the model can follow

Writes the strings the model uses to choose and act: structured tool descriptions, schema text, and sectioned instructions built from runtime context.

Install with Premium View pricing

Plan: Premium
Access: Gated
Install: Tokenized
Authored files: 3

Install with skills CLIPremium skills install through a short-lived private link.

$ npx skills add … --skill write-prompt-contracts

Open Premium install

Filesystem

Browse the authored file tree — tools, references, and evals included. Source unlocks with Premium.

SKILL.mdmarkdown

Unlock skill source

Browse every authored SKILL.md with premium.

Install with Premium

File tree for write-prompt-contracts stays visible so you can see the pack before unlocking.

File

What you get

Descriptions that route

Six-part tool contracts: what it does, when to use it, when not to, hard boundaries, usage caps, and examples.

Agency instead of a name tag

Sectioned instructions that tell the agent to use tools and answer, instead of explaining what it would do.

Gravity named out loud

Find the most general tool the model defaults to, then make every sibling refuse that job by name.

Prompts from runtime context

Build instructions from the tools that are actually wired, so a subagent does not inherit the full catalog.

Where this skill lands

Wording that routes and acts: description contracts, Agency, and instructions derived from wired tools.

A support agent that used search_help for every 'where's my order' question.
A RAG agent that narrated a retrieval plan and waited for approval it did not need.
A commerce subagent that only has catalog tools but inherited the checkout prompt.
Overlapping lookup and search tools with one-way redirects that still leaked.

Write the strings the model actually uses

Available with Premium. The private install link expires; the downloaded skill does not. Generate a fresh link when you want to update it.