Explore Fray

A skill for intelligence that compounds

Useful work should make future work more capable.

Fray stitches Ought, Possible, Deeper, Turnless, and Further into the way an agent works—so it can build on what is known, keep moving, and leave more behind.

FrayStitching skill
01$Oughtchoose
02$Possiblebegin
03$Deeperlearn
04$Turnlesscontinue
05$Furtherpreserve
FurtherPossiblefor what comes next

What gets lost

An answer can survive while the ability to produce it disappears.

Notes and documents may remain after a task ends. The method, judgment, and learning behind them usually do not remain in a form another agent can use.

OutputSaved
MethodUsually lost
JudgmentUsually lost
LearningUsually lost

The principle

Capability—not information—is the unit of compounding intelligence.

Informationrecords what happened.

Capabilitychanges what can happen next.

How Fray works

One skill. Five enduring questions.

Fray brings the concepts into the work when they are useful. They are not steps, and an agent can revisit any of them as conditions change.

What deserves attention?

Choose what is worth doing.

Ought gives finite attention a direction. It names the outcome, the boundary, and what would make the work worth trusting.

Explore ought.sh

How work compounds

What one task learns becomes the next task's starting point.

Further preserves useful capability. Possible makes it available when new work begins. Fray keeps that handoff connected.

Work nowUseful task completed
  • Method
  • Judgment
  • Learning
$Further
RetainedReusable capability

Validated and ready to build upon.

$Possible
Work nextA higher starting point

No reconstruction required.

Begin further ahead

The next agent should never start from zero.

It should begin with the capabilities previous work created—and leave future work more capable in return.

$Ought$Possible$Deeper$Turnless$Further