# Smart Shot

A short request can contain several jobs the person writing it cannot yet see. It may cross product, engineering, security, legal, commercial or operational boundaries. Starting with the obvious interpretation can produce an excellent answer to the wrong problem.

Smart Shot discovers the real work before it staffs the work. It maps explicit and hidden intents, affected domains, dependencies and unknown unknowns. It then assembles ad hoc expert agents from those findings, gives them suitable tools, integrates their work, performs authorized actions and uses independent acceptance where the result needs it.

## Use it when

- The prompt is short, ambiguous or consequential.
- Words such as best, complete or high quality hide a real comparison.
- You do not yet know which professional domains or experts the work needs.
- Missing dependencies or unknown unknowns could change the result.
- The work may require evidence, tools, authorized action and independent review.

## Example requests

Smart Shot can start from two words:

```text
Analyze incident.io
```

```text
What are the best options for preserving useful website analytics?
```

```text
How should we organize company knowledge so agents can use it without exposing private information?
```

## What to expect

Smart Shot costs more time and model work than Fast Shot. In return, it discovers the work that the prompt left unsaid, creates the specialist team at runtime and tests the integrated result instead of trusting a polished answer.

## Install and source

```shell
npx skills@latest add simonasrazm/skills --skill smart-shot
```

Read the [Smart Shot source](https://github.com/simonasrazm/skills/tree/main/skills/smart-shot) or the [skills.sh listing](https://www.skills.sh/simonasrazm/skills/smart-shot).

Source: https://finallydone.ai/skills/smart-shot/