Start with the exact profile and error

Record the command, selected profile, plugin source, and first meaningful error before changing anything. Many apparent plugin failures are profile mismatches: the package was added to one profile while a different profile was launched.

  • Confirm the profile name in both the install and launch commands.
  • Copy the first missing package, service, or configuration error.
  • Check the plugin README for supported dsh versions and prerequisites.

Check profile dependency state

List the dependencies managed inside the affected profile. If a transitive package is involved, use why to see which direct dependency brought it into the profile.

npx -y @deepseek-ai/dsh plugin --profile web ls
npx -y @deepseek-ai/dsh plugin --profile web why <package>

Inspect the composed plugin tree

Dumping configuration helps separate an installation problem from a composition problem. If the package is installed but its expected bundle layer is missing, verify that its package manifest declares a dsh bundle and that you installed the documented build or release.

npx -y @deepseek-ai/dsh --profile web --dump-default-config
npx -y @deepseek-ai/dsh --profile web --dump-config

Remove one suspect plugin

When the profile stopped booting immediately after a plugin change, remove that single package through the same profile-scoped command. Avoid deleting the entire Harness home because that also removes unrelated profile state and makes the original cause harder to prove.

npx -y @deepseek-ai/dsh plugin --profile web remove <package>

Prepare a useful issue report

If the failure remains, reduce it to the smallest reproducible profile and report the installed dsh version, operating system, Node.js version, exact plugin source, commands, and error output. Redact credentials and private workspace paths.

A community plugin can lag behind a developer-preview API. Pinning a known release is often safer than repeatedly reinstalling the newest commit.