PRODUCT DEFINITION / POSSIBLE.SH
Possible is the
outcome layer
for AI agents.
Describe the result you want. Possible clarifies the outcome, recommends a reviewed pack, assembles specialist skills, and coordinates the work until there is evidence it is done.
One ambition. A coordinated path. A verified result.
- 01Your ambition
You explain what you want to make possible in ordinary language.
- 02Possible
Clarifies the intended end state and recommends the appropriate pack.
- 03Pack + specialists
Defines and performs the work after you approve the recommendation.
- 04Verified outcome
Artifacts, checks, limitations, and receipts show what is actually true.
What each thing means.
Outcome
The observable end state you want to make true—not the activity an agent performs.
Outcome pack
A reviewed way to reach a class of outcomes, including its workstreams, ingredients, boundaries, and checks.
Specialist skills
Focused abilities used as ingredients. They do the design, coding, testing, CAD, release, or operational work.
Possible
The skill that clarifies the outcome, recommends the pack, coordinates the agents, and keeps the work attached to proof.
The website is the public home, not the agent.
Possible.sh hosts the installer, the reviewed pack catalog, documentation, demonstrations, and evidence. The installed Possible skill is what joins you inside the project and coordinates the work.
Read the documentation →- Install
npx @fraylabs/possible init- Interface
$possibleand a conversation- Knowledge
- Reviewed outcome packs
- Execution
- Codex and specialist skills
- Result
- A verified outcome
What Possible is not.
- Not another directory where you must choose individual skills.
- Not one giant skill that can perform every kind of work itself.
- Not an outcome—the outcome is the result you are trying to achieve.
- Not blanket permission to deploy, publish, spend, contact people, or change external systems.
What do you want to make possible?
npx @fraylabs/possible initThen open Codex and type $possible.