Skill Detail

Diagnose Paradev host and terminal failures with the doctor skill

Capture Paradev service, account, host and version evidence, propose one symptom-matched repair, and verify the original failure before escalation.

Runbooks & DiagnosticsMulti-Framework
Runbooks & Diagnostics Multi-Framework Security Reviewed
⭐ 15.6k GitHub stars
COPY SKILL INSTRUCTIONS (OPTIONAL)
npx skills add agentskillexchange/skills --skill diagnose-paradev-host-and-terminal-failures-with-the-doctor-skill Copy
Uses the third-party skills CLI, not an ASE-owned installer. Check your agent’s compatibility. This copies instructions; complete the upstream tool setup below separately.
At a glance
Tools required
Paradev desktop app and superset CLI (paradev alias); shell access; existing SSH access only for an affected remote host; optional relevant local service logs
Install & setup
Install the Paradev desktop app from https://github.com/ParadevAI/Paradev/releases/latest and launch it; upstream documents automatic managed skill provisioning at launch. Verify the bundled CLI with superset –version. In Claude Code invoke /superset:doctor with the symptom; other supported agents discover the managed skill under the superset- prefix. Source reviewed at revision 7b3643f33143bd0d479a90f53125651803b9961d. No upstream installation or authenticated runtime test was performed during metadata review.
Author
ParadevAI
Publisher
Organization
Last updated
Oct 27, 2026
Quick brief

Use this procedure when Paradev, formerly Superset, cannot connect to a host, attach a terminal, authenticate an existing session, or reconcile desktop and CLI versions. It turns a reported symptom into a diagnostic snapshot, one proposed repair, and a repeat of the failing action. It is based on the upstream doctor skill, rather than a general introduction to the IDE.

How it works

What this skill actually does

When to use

Prerequisites and invocation

Use an existing Paradev installation and its superset CLI; paradev is an alias for the same commands. The desktop app provisions its managed skills at launch. Describe the failing action and affected host to the agent. In Claude Code, explicitly invoke /superset:doctor; other supported agents discover the managed skill with the superset- prefix. Confirm the active organization before diagnosing a remote host.

Capture a read-only snapshot

superset status
superset auth whoami
superset hosts list
superset --version

Keep the result or error from each command, even when another command fails. Compare service health, account and organization, host reachability, and CLI version with the reported symptom. Record the original failing action so the same action can be used for verification.

Match evidence and propose one repair

An expired session points to superset auth login; an offline known host points to superset hosts wake <id> and checking that the machine is awake and online. A stopped host service points to superset start. A stale CLI or desktop mismatch points to superset update. These are proposed repairs: obtain the operator’s go-ahead before running any command that changes state. Diagnose one cause and verify one change at a time.

For a host previously configured with hosts add, first confirm existing SSH access, the remote operating system, and curl availability. Upstream supports macOS or glibc Linux on x64 or arm64 for that setup path. If the evidence calls for repairing that installation, propose re-running superset hosts add <ssh-target> as a separate approved action. Do not turn diagnosis into new host onboarding. On Linux, the relevant service log is available through journalctl –user -u superset-host -n 50 on that host. On macOS, app-side errors may be inspected in the newest entries of ~/Library/Logs/Superset/main.log. Inspect only relevant excerpts and redact private context.

Verify and hand off

After an authorized repair, repeat the original failing action. Report the observed failure, supporting snapshot, exact approved change, and verification result. If unresolved, return a concise diagnostic packet and offer escalation through the upstream feedback skill. Obtain permission before including logs or filing feedback, because logs can contain paths and project names.

Scope boundary

This skill diagnoses Paradev itself. It does not dispatch coding workers, build orchestration plans, set up a new fleet, delete application data, reset machine identity, or publish a support report automatically. Read-only findings remain useful when no repair is authorized. Metadata verification confirms source and invocation documentation; it does not certify security or prove that a repair succeeds in a particular installation.