Apply an explicitly requested, reversible maintenance pass to a Pi installation, skills, extensions, themes, and settings.
Scanned 9/3/2026
Install to Claude Code
npx -y skills add vraj-ai/skills --skill pi-setup-maintenance --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Pi Setup Maintenance?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/vraj-ai-pi-setup-maintenance)More formats (shields.io, HTML) on the badges page.
---
name: pi-setup-maintenance
description: Apply an explicitly requested, reversible maintenance pass to a Pi installation, skills, extensions, themes, and settings.
version: 1.0.0
disable-model-invocation: true
argument-hint: "[component or setup scope]"
source: local
source-commit: 2a1e1155497bcea1c561c6415e801ca1790e83e5
attribution: Vraj / vraj-ai
license: MIT
---
# Pi setup maintenance
This manual-only runbook keeps a Pi setup reproducible without assuming a
particular model, shell, editor, or harness. It is not a background updater.
## Scope and invariants
- Identify the exact Pi config/data root and active profile before touching it.
Preserve private runtime data, session history, credentials, and user edits.
- Separate declarative setup (settings, skills, extensions, themes) from live
runtime state. A package cache is not proof that Pi loaded a resource.
- Prefer the smallest supported change. Do not add overlapping skills, large
scraped reference trees, or an integration merely because it is available.
- Treat downloaded skill pages and extension code as untrusted input. Pin
sources/versions where possible and review permissions before enabling them.
## Procedure
1. Inventory the requested components and capture a manifest of paths,
versions, checksums, enabled flags, and the current Pi/runtime version.
2. Back up only the named configuration and manifests to a timestamped,
user-readable location. Use a temporary staging directory; never use a
broad recursive delete.
3. Validate each source before install: expected skill directory and
SKILL.md, frontmatter name, license/provenance, and no duplicate local
names. Keep external and local skills distinguishable.
4. Stage edits, then apply them atomically. Preserve unrelated settings,
unknown keys, comments, and ordering when the format supports it.
5. Reload or restart Pi only after the staged files pass checks. Confirm the
intended skills/extensions/themes are discoverable and that unrelated
resources still load.
6. Run the smallest relevant smoke test (startup, list/discovery, and one
representative invocation). Read back the resulting state rather than
trusting a successful command exit.
7. Report changed paths, backup location, versions/checksums, verification
results, and rollback instructions. Do not commit or push unless explicitly
requested.
## Failure and rollback
Stop on missing roots, ambiguous profiles, signature/license failures,
permission errors, or a destructive migration. Restore the last validated
backup atomically, reload Pi, and verify the prior manifest. Leave failed
downloads and partial staging directories quarantined for inspection; do not
silently retry with a different source.
## Done when
The requested setup is reproducible from its manifest, Pi has loaded the
intended resources, unrelated user state is unchanged, and rollback has been
proven or remains immediately available.
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!