agentsclimarketplace

Setup rootloom

Skill liyanqing90/rootloom/plugins/rootloom/skills/setup-rootloom

Plan, install, inspect, update, or roll back Rootloom Personal Core in a user's Codex home. Supports Skills-only, guidance, and the recommended personal preset. Use when the user explicitly asks for a Rootloom setup plan, installation, configuration, bootstrap, repair, audit, status, update, reduction, rollback, or removal. Never overwrite user-owned files without showing the plan and obtaining exact replacement authorization.From its SKILL.md

Install
npx -y skills add liyanqing90/rootloom --skill setup-rootloom

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 7 stars7 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.4 KB, ~1.0k tokens by cl100k_base, as published. Nobody here has run it

Set up Rootloom Personal Core

Resolve this Skill directory and use its deterministic setup script.

Choose a preset

python3 <skill-dir>/scripts/setup_rootloom.py list-components
PresetResult
skills-onlyPlugin Skills only; the project-guidance Hook is disabled
guidanceGlobal working agreement plus read-only project-context Hook
personalGuidance plus optional Autonomy; recommended and default

Exact capabilities are global-policy, project-context, and autonomy. autonomy always includes global-policy because the Rules intentionally defer authorization mode and scope to that guidance. Legacy engineering and command-safety inputs remain compatibility aliases but are not recommended names.

Optional global layer

codex plugin add rootloom@rootloom completes the plugin installation. It exposes Rootloom Skills without writing global guidance, command Rules, setup state, or review gates. Stop there for the lowest-cost experience.

Only when the user explicitly wants cross-project guidance, the project-guidance Hook, or command Rules, inspect and install an optional preset:

python3 <skill-dir>/scripts/setup_rootloom.py plan --preset personal
python3 <skill-dir>/scripts/setup_rootloom.py install --preset personal
python3 <skill-dir>/scripts/setup_rootloom.py status

apply remains a compatibility/expert command. Prefer install for the first setup and upgrade after the Codex plugin snapshot has been refreshed.

The personal preset manages only:

  • ~/.codex/AGENTS.md;
  • ~/.codex/rules/rootloom.rules;
  • ~/.codex/.rootloom/components.json;
  • ~/.codex/.rootloom/state.json and a simple backup manifest.

Setup is serialized by an ordinary local lock, refuses symlinked or user-owned targets, backs up every replaced file before mutation, writes each file atomically, and preserves original content and mode for rollback. It does not claim whole-transaction crash compensation or hostile shared-filesystem locking.

If the plan contains conflict, show the exact paths. Use --replace-conflicts only after the user authorizes those replacements. A backup does not replace authorization.

Verify optional Autonomy

When autonomy is selected, run codex execpolicy check for at least:

  • routine and high-risk commands represented in the Rules → allow, because static argv Rules cannot carry authorization-mode state;
  • catastrophic recursive deletion of root, home, the current directory, or its parent → forbidden.

The installed global guidance owns three authorization modes. Single action covers one displayed command/action. Standard is the persistent cross-task default for all non-high-risk steps required by each explicit goal; every task still resolves its own operation type and scope. Full covers high-risk steps only in the current task and scope and is never inferred. When a Standard task reaches a high-risk boundary, offer all three modes once. Rules avoid duplicating that semantic decision; they do not grant authority themselves. Other active Rules and Codex platform policy remain authoritative and may still require approval.

Start a new Codex task after setup or plugin update so assets and Hooks are rediscovered.

Upgrade

Refresh and reinstall the plugin with Codex, then start a new task. That completes the normal plugin upgrade and does not run analyzers, contracts, finalizers, or setup validation:

codex plugin marketplace upgrade rootloom
codex plugin add rootloom@rootloom

If the user previously installed the optional global layer and wants its copied assets refreshed, run one explicit command:

python3 <skill-dir>/scripts/setup_rootloom.py upgrade

Upgrade preserves the installed capability selection. It reports up_to_date for the same version, records a version-only upgrade without a redundant asset backup, and backs up changed managed assets before replacement. A pristine target retired by the new catalog is backed up and removed, so rollback restores it; a drifted retired target is refused. All installed target paths are revalidated before access. If any managed target drifted after setup, both status and upgrade expose/refuse it; restore the installed content or roll back instead of overwriting the edit. --replace-conflicts applies only to newly introduced user-owned targets after exact authorization, not post-install drift.

Roll back or change preset

python3 <skill-dir>/scripts/setup_rootloom.py rollback

Rollback first validates that managed files still match the installed hashes. It refuses to overwrite post-setup edits, then restores the backed-up content and mode or removes files created by setup. A normal rollback returns to the previous simple backup; rollback --all follows the available backup chain to the pre-install state.

To change presets, roll back, plan the new selection, and apply it. Archived Assurance Edition 1.2.19 uses a different branch and installation contract; roll it back with that version before installing Personal Core.

What ships with it: 3 files

31.3 KB alongside SKILL.md, 1 of them executable

agents/

scripts/

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.