Claude Skills by imran31415

Safely edit the workspace pod's boot scripts (start.sh, the entrypoint/ssh-server configmaps) in kube-coder. Use when changing anything that runs at pod start — credential wiring, service launch, tmux/ttyd config, symlinks — where the two-home layout and Helm templating have non-obvious footguns.

Spin up an isolated agent to work a kube-coder issue. Given an issue number, creates a clean git worktree branched from a freshly-fetched origin/main, pulls the issue text, and launches a background Claude task that is BORN inside that worktree (workdir=worktree) so it cannot work in the shared clone. Use whenever the user wants to "work on issue N", "start an agent on issue N", or set up a per-issue worktree. Also lists issue worktrees and their task status.

Run kube-coder's full CI suite locally before pushing — helm lint + unit tests, server.py tests, dashboard + controller SPA builds + vitest, and shell syntax checks. Use before opening or updating a PR so CI is green on the first push.

Scope a kube-coder change before you open or update a PR — what the diff actually touches, which tests reach it, what the change made worse, and what is left untested. Use after the code is written and before kc-preflight/kc-ship-pr, so the PR description and the tests you run come from the call graph rather than from memory. Also the answer to "is this diff bigger than I think?" and "which tests should I run for this?".

Capture desktop + mobile, dark + light screenshots of the kube-coder dashboard SPA for visual QA of a UI change. Use when the user wants to see a frontend change rendered in a real browser rather than just unit-tested.

Commit local changes and open a pull request against kube-coder from inside a workspace pod. Use when the user wants to push a branch or open/update a PR with the workspace GitHub App token.

Launch a Claude task on a remote kube-coder workspace, check task status, or attach to a running session. Use when the user wants to send work to a remote workspace pod.

Create an isolated git worktree + branch + port for this session so concurrent kube-coder sessions and builds don't conflict. Use at the START of any change you want isolated, when two sessions touch the same repo, or before running a dev server/build that would collide on ports or output dirs. Also lists and tears down worktrees.