← ClaudeAtlas

temperlisted

Testing umbrella: runs ring-test/wield/proof/hone, one READY|NOT READY verdict. Triggers: 'temper', 'test everything'.
dyasnurhakim/smithy-claude · ★ 0 · Testing & QA · score 67
Install: claude install-skill dyasnurhakim/smithy-claude
# Temper — Testing Umbrella (Tempering = controlled stress until the metal holds.) Read `${CLAUDE_PLUGIN_ROOT}/references/creed.md`, `${CLAUDE_PLUGIN_ROOT}/references/memory.md`, and `${CLAUDE_PLUGIN_ROOT}/references/stacks.md` first. Log: `bash ${CLAUDE_PLUGIN_ROOT}/scripts/ledger.sh append temper <slug> suite STARTED -` ## Process 1. **Detect the stack:** `bash ${CLAUDE_PLUGIN_ROOT}/scripts/stack-detect.sh`. Show the result. If `stack=unknown` or it contradicts the repo, confirm with the user before proceeding. 2. **Select the suites.** Read `testing.skip` from the effective config: `bash ${CLAUDE_PLUGIN_ROOT}/scripts/config.sh get testing.skip` — it merges all three layers (defaults → global → project), so never read a config file directly or you will miss the user's global default. Then AskUserQuestion (multiSelect) with the remaining suites, defaults pre-picked: - ring-test — always suggested - wield — always suggested - proof — suggested ONLY if a runnable service exists (spec/STATE/ask) - hone — suggested if the job touched hot paths or the user cares about perf Note skipped suites in the summary — a skipped suite is a gap, not a pass. 3. **Run the selected skills in order** ring-test → wield → proof → hone, each per its own SKILL.md (each writes its own report + ledger line). A FAIL does not abort the remaining suites (full information first) — EXCEPT: skip proof if wield found a Critical (don't load-test a broken app).