Why it Matters
Agent skills are not programmatic pieces of code that dictate what an agent can or can’t do. They are additional context injected into the agent session with the purpose of telling the agent what it should do and how it should do it. But agents often just infer that context or skip parts of the skill. Skills then are just guidelines that may or may not be strictly adhered to.
The 3-factor invocation method is a way to instruct the agent as clearly and cleanly as possible to follow the skill exactly as you intend it to be followed.
How it works
The method involves three steps:
01
Invoke
At-mention your skill directly in your prompt. This ensures the Agent goes there first.
02
Goal
State your goal clearly and make sure it’s aligned with the purpose of the skill itself.
03
Alignment
Ensure your prompt is aligned with the goal of the skill. Without alignment the Agent might just infer.
Example in Practice
Here’s some examples taken from the @robots-txt-audit skill:
Leverage @robots-txt-audit on mattcromwell.com with max_visibility as the goal
This works because it does exactly what the method describes:
- At-mentions the skill explicitly
- States the goal clearly
- And the goal of the prompt matches the goal of the skill exactly
Common Mistakes
The biggest mistake is just saying “Read my robots.txt file and tell me what to change.” AI agents are very intelligent and thrive with a wide scope of work to run with. But with a prompt like that it’s highly likely that the skill you created will be ignored completely or just lightly referenced. The guidance you’ll get will be inferred from the model’s general context which is informed by the entire web, and not by the expertise you hand-crafted for it.
