If you already think in Skills, you already understand Decorators. One gives your agents reusable expertise. The other gives your prompts reusable shape. Decorators are a deterministic way to level up prompts — you spend time on intent, not on syntax.
A decorator is a packaged prompt instruction. A hook expands it into the careful version before the model ever sees your prompt. Here’s what that actually looks like.
You type:
::Antipatterns(domain=database, severity=critical, format=examples)
Review my MongoDB schema design.
The model receives:
Identify common antipatterns and mistakes in the specified domain. Focus on how to recognize these issues and provide guidance on how to avoid or fix them. Focus on database antipatterns such as poor schema design, inefficient queries, and data integrity issues. Focus only on the most severe antipatterns that pose immediate risks or significant technical debt. Provide concrete examples of each antipattern with code or design snippets that illustrate the problem.
Review my MongoDB schema design.
One line in. A five-sentence specification out — domain narrowed, severity filtered, output format dictated. The model now performs a fundamentally different review than it would for the bare prompt. You wrote the sigil once. The engine assembles the rest, the same way, every time.
Same pattern as Skills, with one important difference. Skills are loaded into context and the model decides whether to use them — they’re advisory. Decorators run in a UserPromptSubmit hook before the model ever sees your prompt — they’re deterministic. Use one, get its shaping, every time, no exceptions.
Turn on auto-decorate and you don’t even pick.
Install:
claude plugin marketplace add synaptiai/synapti-marketplace
claude plugin install prompt-decorators
Neither "product engineer" nor "AI engineer" describes what a growing number of people actually do: take an idea from inception to production using AI as their primary collaborator across the entire lifecycle.
That's the AI product engineer.
Someone who does the work — hands on, with AI — across strategy, specs, code, testing, evals, and operations. Pressure-testing a product narrative in the morning, co-developing the implementation in the afternoon, generating adversarial test cases before dinner.
Four things define this role:
. -- . The full loop from idea to production behavior — on systems where "works" is statistical, not binary.
. - . AI in every stage of building, not just in the product. If you can't use AI to accelerate correctly across the lifecycle, you're doing a slower job with a new title.
. . Making reliability measurable on systems that behave differently every run.
. . Models shift, tools evolve. The job is to experiment fast, adopt without chaos, and converge under uncertainty. Curiosity isn't a personality trait — it's a production skill with receipts.
This changes how you hire. Past titles tell you what someone did in a world that no longer exists. The question is: can they learn fast and ship under uncertainty with tools that didn't exist six months ago?
Agent Capability Standard is an open specification for composable AI agent capabilities. It defines 36 atomic capabilities across 9 cognitive layers, a type-safe workflow DSL, and grounded world modeling with trust-aware conflict resolution. Built on the Grounded Agency philosophy, it makes agent reliability structural—not optional.
A decorator is a packaged prompt instruction. A hook expands it into the careful version before the model ever sees your prompt. Here’s what that actually looks like. You type: ::Antipatterns(domain=database, severity=critical, format=examples) Review my MongoDB schema design.
The model receives: Identify common antipatterns and mistakes in the specified domain. Focus on how to recognize these issues and provide guidance on how to avoid or fix them. Focus on database antipatterns such as poor schema design, inefficient queries, and data integrity issues. Focus only on the most severe antipatterns that pose immediate risks or significant technical debt. Provide concrete examples of each antipattern with code or design snippets that illustrate the problem. Review my MongoDB schema design.
One line in. A five-sentence specification out — domain narrowed, severity filtered, output format dictated. The model now performs a fundamentally different review than it would for the bare prompt. You wrote the sigil once. The engine assembles the rest, the same way, every time.
Same pattern as Skills, with one important difference. Skills are loaded into context and the model decides whether to use them — they’re advisory. Decorators run in a UserPromptSubmit hook before the model ever sees your prompt — they’re deterministic. Use one, get its shaping, every time, no exceptions.
Turn on auto-decorate and you don’t even pick.
Install: claude plugin marketplace add synaptiai/synapti-marketplace claude plugin install prompt-decorators
Plugin marketplace: https://github.com/synaptiai/synapti-marketplace See all available decorators: https://synaptiai.github.io/prompt-decorators/api/decorators...