When the user needs claude-arsenal/ set up in a host repo, or wants to register a workspace via --workspace. Re-running is safe (refreshes stale bundle files only). Do NOT use to add tasks (see queue-add) or resume the worker loop (see continue).
Use whenever anything under src/core/parsers/ is being added or repaired — capturing a live DOM, deriving selectors, implementing a BaseParser subclass, and verifying live that every widget the chatbot renders survives into an export. Triggered by "add a…
Use whenever anything under src/core/exporters/ is being added or repaired — a new output format, a content type missing from some formats, or an exported file that will not open. Triggered by "add EPUB/CSV export", "add a new export format", "images are…
When the user wants to resume work or run the worker loop — picks the next unblocked task, optionally scoped by tag(s) and/or a workspace, or matched by title text. Use /continue [TAG … | WORKSPACE | search-text]. Do NOT use before running init.
Use whenever the user wants to turn a coverage.py report into a ranked list of the highest-value missing tests — runs analyze_coverage.py over coverage.json and surfaces contiguous missing-line runs marking whole untested functions or branches. Triggers —…
Use whenever a uv-managed Python project's dependencies need upgrading safely — the uv lock --upgrade, pip-audit (CVEs), test-gate, classify-breakage runbook. Runs compare_lockfile.py to split lockfile churn into direct vs transitive before a transitive bump…
When the user is defining the technical solution after discovery — contracts, task split, risk register, sequencing. Owns scripts — validate_plan. Do NOT use for problem investigation (see specify), implementation (see execution), or PR review (see review).
When the user is implementing code changes from a design — code change, tests, merge-ready output. Do NOT use for investigation (see specify), design (see design), or routine one-off scripts that bypass the design step.