# Intelligence Upgrade

> Upgrade Intelligence and installed packages with migration checks

- Skill: `ainova-systems/intelligence-upgrade` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ainova-systems/intelligence-upgrade`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ainova-systems/intelligence-upgrade/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: ainova-systems (https://skillmd.com/u/ainova-systems)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ainova-systems/intelligence-upgrade

---


# Upgrade Intelligence and packages

The CLI owns planning and application. This skill interprets the plan, reads
the changelog across an engine-version gap, obtains approval, and verifies the
result.

## Steps

1. Run `intelligence update --preview` (or
   `intelligence update $ARGUMENTS --preview` for one named package). Use its
   CLI, project, and package sections as the complete plan; do not re-resolve
   versions independently.

2. If the CLI or project engine version would change, read the authoritative
   `CHANGELOG.md` at `https://github.com/ainova-systems/intelligence` for every
   release in **`current < release <= target`**. If it is unavailable, stop
   before changing versions. Turn every crossed `### Breaking` item into a
   post-condition to verify.

3. Show the plan and breaking checklist to the user. The command without a
   mode also shows the plan and prompts; after approval, use `--apply` for an
   unambiguous non-interactive execution.

4. If the plan reports a newer global CLI, run `intelligence upgrade --apply`
   after approval — it installs exactly the version the plan showed — then
   rerun `intelligence update --preview` with the new executable. When the
   project section reports a schema stamped ahead of the stable line, the CLI
   that stamped it was a prerelease: use `intelligence upgrade --next --apply`.
   If `upgrade` refuses because npm did not make this installation, run the
   command its message names instead. Apply the resulting plan with
   `intelligence update --apply`, or `intelligence update $ARGUMENTS --apply`
   when one package was requested.

5. An applied update that renders must finish with `IS_STATUS=ok`; preserve and
   stop on any other status. Then run `intelligence status --check`. Do not run
   a duplicate sync after `update --apply`. Verify every crossed breaking
   post-condition directly and report any item that cannot be machine-verified.

Report versions before and after, the applied plan, post-condition results,
and any remaining action. On a refusal, preserve the full error and stop
instead of invoking hidden lifecycle operations.