Skill Authoring
Today Jumpspace installs a reference workflow skill and named pipeline skills for supported agents.
Current named skills package narrower workflows:
| Skill | Use when |
|---|---|
jumpspace-bootstrap | Advanced import from existing docs into task memory. |
jumpspace-work | Advanced implementation from a ready task. |
jumpspace-review | Reviewing drift and preparing PR output. |
jumpspace-intent-review | Local review when changed files match active intents. |
jumpspace-handoff | Switching agents or preserving session state. |
Install one named skill for one supported agent:
npx @jumpspace/cli add-skill jumpspace-work --agent claudenpx @jumpspace/cli add-skill intent-review --agent codexnpx @jumpspace/cli add-skill bootstrap --agent codexNamed installs always include the reference jumpspace-workflow skill too.
Authoring Principles
Section titled “Authoring Principles”Good Jumpspace skills should:
- start from source-backed evidence
- ask for human approval before mutation
- avoid treating retrieval as an answer
- prefer JSON contracts for scripting
- record durable evidence before handoff
- tell agents when to fall back to code search
Current Command To Install Guidance
Section titled “Current Command To Install Guidance”npx @jumpspace/cli add-skill --all