FAQ
Common questions, short answers, links to deeper detail.
What's the difference between rig and tap?
Rig is the workspace product: the CLI, the manifest, the artifact format,
the Hub. Tap is the sync system that powers
shared workspaces. You interact with it through the
rig workspace verbs (rig share, rig status, rig resume); tap is
plumbing.
Long answer: How collab works.
Do I need tap to use rig?
No. Solo rigs work without tap. Only install @rigxyz/tapd if you're using
shared rigs.
Can I publish a rig privately?
Yes, three ways:
- Local file registry on a shared drive or NFS mount.
- GitHub release on a public repo (effectively public, but discoverable only by URL).
- Just hand someone the
.rig.tgzdirectly —rig hub use ./path.rig.tgzworks.
Private Hub publishing is on the roadmap; not yet available.
Does rig install MCP servers / Python / Node for me?
No. Rig declares what's required and reports what's missing, but it doesn't install language runtimes or MCP server packages. That's the user's job (possibly with the agent's help).
This is by design — see Concepts.
Why don't my data files ship?
By default, data/portfolios/, data/transcripts/, outputs/,
private/, research/ are excluded from the artifact. The reasoning is
that they typically contain personal or user-specific data. To opt files
in, use rig add. See What gets shared.
Why doesn't rig pack include .env?
Because it would ship your secrets. .env and .env.* are in the default
excludes. Declare your env vars in
[env] and ship a scripts/.env.example
template instead.
Does the Hub host my secrets if I forgot to exclude .env?
The Hub itself doesn't strip the artifact. If you accidentally include .env,
it'll be visible in the artifact. Always rig inspect
before publishing.
If you discover you've published a secret, rotate the secret immediately and contact the Hub admin to take down the artifact.
What's the difference between rig use and rig install?
rig use downloads and extracts the artifact.
rig install is the side-effecting setup step (env vars,
dirs, permissions). They're split so the destructive step is explicit.
Can two people on different machines edit the same rig?
Yes — with sync. The owner adds [sync]
to rig.toml, runs rig share --enable, and invites collaborators. Edits
sync in near-real-time.
What if I delete a file by accident in a shared rig?
Recovery. rig history <path> shows
every event; rig restore <path> --as-of <cursor> brings it back.
Can I publish a rig under someone else's author?
No. The Hub checks the manifest's [rig].author against your authenticated
user and rejects mismatches.
Why is rig hub publish refusing on my synced rig?
Sync is dirty. See Publishing a synced rig.
Get rig status to all green, then retry.
What permissions does rig grant Claude?
Only what the rig declares in [permissions],
merged with your existing .claude/settings.json (union for allow, union
for deny, deny wins). Rig can ask for narrow permissions to skip dialogs;
it can't weaken your own deny rules.
Does rig send telemetry?
No. The only network traffic is to whatever source you rig use, the Hub
(when you rig hub publish, rig hub search, rig login), and (for collab) the
sync relay.
What happens if I publish a rig and then change my mind?
Versions are immutable once published. Bump and republish, or contact the Hub admin for takedown.
Can I use rig with agent runtimes other than Claude Code?
Not yet. [rig].runtime is restricted to "claude-code" for now. Other
runtimes are on the roadmap when their permission/skill models stabilise.
Where's the CLI source?
Open source on GitHub. (Link from the home page.)
How do I report a bug?
GitHub issues on the rig repo, or email the Hub admin (linked on userig.xyz).