Skip to content
Jumpspace

Agent Skills

Jumpspace is most useful when agents know to use it first, then fall back to code reading when needed.

Install Skills

Section titled “Install Skills”
Terminal window
npx @jumpspace/cli add-skill --codex
npx @jumpspace/cli add-skill --claude
npx @jumpspace/cli add-skill --all
npx @jumpspace/cli add-skill jumpspace-work --agent claude

The installer is additive and non-destructive. It creates missing files or updates clearly marked Jumpspace-managed blocks. It should not overwrite user-authored guidance outside those markers. Named installs add the reference workflow skill plus the selected pipeline skill.

What The Skill Tells Agents

Section titled “What The Skill Tells Agents”

The generated guidance tells agents to:

  • run Jumpspace discovery before broad repository reads
  • treat task ask as evidence, not authority
  • use JSON schemas before scripting against command output
  • use task work for approved implementation
  • use task next for plan steps
  • record step evidence
  • use task verify to earn verified state
  • run task scan, task audit, and task doctor after metadata mutations
  • fall back to rg and direct code reads when Jumpspace evidence is missing

Useful Agent Commands

Section titled “Useful Agent Commands”
Terminal window
npx @jumpspace/cli task find "approval flow" --json --compact
npx @jumpspace/cli task context DOC-EXAMPLE-001 --json
npx @jumpspace/cli task work DOC-EXAMPLE-001 --json
npx @jumpspace/cli task next DOC-EXAMPLE-001 --json
npx @jumpspace/cli task handoff --task DOC-EXAMPLE-001 --json
npx @jumpspace/cli task last --json
npx @jumpspace/cli task history --json

Check The Active Install

Section titled “Check The Active Install”

When an agent seems to be running an old CLI, run:

Terminal window
npx @jumpspace/cli release install-doctor --json

This reports the invoked binary, the first jumpspace found on PATH, package roots, versions, schema counts, workspace comparisons, stale-install warnings, and repair commands.

Human Pattern

Section titled “Human Pattern”

For a small team, the ideal pattern is:

  1. The human writes or approves the task block.
  2. The agent uses Jumpspace to gather context.
  3. The human approves the plan.
  4. The agent implements the next step.
  5. Jumpspace records evidence and verification.
  6. CI checks drift before merge.