# Make skills durable.

> Turn reusable procedures and instructions into versioned, inspectable capabilities with provenance, dependencies, activation conditions, and target-aware compatibility.

Canonical: https://rangoon.ai/product/skills/

Status: In active development. Product screenshots are design previews with illustrative data. No general availability is implied.

## Treat instructions as assets

A planned registry records identifiers, content digests, owners, inputs, outputs, tools, references, tests, and rollout state.

### Lifecycle clarity

Draft, validate, test, review, publish, deploy, deprecate, and archive are distinct stages.

## Merge without erasing source

Compare overlap, conflicts, shared procedures, and dependencies before choosing a composite, shared-core, sequential, or conditional design.

### Preserved ancestry

Original assets and proposed relationships remain traceable through review.

## Split for a reason

Decomposition proposals can separate always-on project context from focused procedures, tool requirements, or policy-sensitive work.

### Human review stays central

Suggested classifications and splits are evidence-backed proposals, not automatic truth.


## More information

- [Documentation index](https://rangoon.ai/llms.txt)
- [AI access and policies](https://rangoon.ai/ai/)
- [Source repository](https://github.com/hypler-dev/rangoon)