← ClaudeAtlas

planlet-completelisted

Validate and safely complete or archive exactly one active repository-local Planlet with a UTC audit record. Use when a user asks to finish the Planlet lifecycle, archive completed work, or explicitly override incomplete tasks with a recorded reason.
vipentti/planlet · ★ 1 · AI & Automation · score 71
Install: claude install-skill vipentti/planlet
# Planlet Complete Complete one planlet without hiding unfinished or invalid work. ## Start the workflow 1. Discover the repository root without traversing above its boundary. 2. Use one available `planlet` executable throughout. Confirm needed operations with `planlet help <command>` and pass `--root "<repository-root>"` to every operational command. Treat angle-bracket runtime values as separate argv values; when invoking through a shell, apply shell-specific escaping instead of interpolating raw text. 3. Resolve exactly one active planlet with `planlet --root "<repository-root>" list`. Accept one valid explicit active slug. With no slug, select and announce sole active planlet; report none; or ask user to choose when several exist. 4. Run `planlet --root "<repository-root>" validate <slug>`. Stop on non-zero exit. Re-read both files completely with `planlet --root "<repository-root>" --full show <slug> --part plan` and `planlet --root "<repository-root>" --full show <slug> --part tasks`; treat missing, unreadable, or malformed files as invalid. 5. Run `planlet --root "<repository-root>" tasks <slug> --remaining`. 6. The `planlet` CLI is required. If no executable is available, install it (`npm install -g @vipentti/planlet`) or invoke it through `npx @vipentti/planlet`. If it still cannot run, stop and report that, naming the missing executable. Do not reimplement CLI operations by editing planlet files. ## Decide completion Inspect all recognized tasks. For n