菜单
Pipeline output quality validation. Generates a real app via /ship, builds it, and grades wiring quality with deterministic checks. Default mode ("quality") generates a recipe app and validates build + wiring. "ship" mode runs the legacy structural pipeline validation.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | smoke-test |
| description | Pipeline output quality validation. Generates a real app via /ship, builds it, and grades wiring quality with deterministic checks. Default mode ("quality") generates a recipe app and validates build + wiring. "ship" mode runs the legacy structural pipeline validation. |
| args | [{"name":"args","type":"string","description":"Test mode to run. \"quality\" (default) generates and validates a real app. \"ship\" runs legacy structural pipeline validation.\n"}] |
User has requested: /smoke-test {{args}}
Inspect {{args}} (trim whitespace):
Usage: /smoke-test [mode]
Available modes:
quality — (default) Generate a recipe app, build it, grade wiring quality
ship — Legacy structural pipeline validation (DB state, plan files)
Create a temp directory and initialize it as a standalone git repo with a minimal Node/TS scaffold:
SMOKE_DIR=$(mktemp -d /tmp/smoke-test-XXXXXX)
cd "$SMOKE_DIR"
git init
git checkout -b main
echo '{"name":"smoke-test-recipe-app","private":true,"scripts":{"dev":"next dev","build":"next build"}}' > package.json
echo '{"compilerOptions":{"target":"es5","lib":["dom","dom.iterable","esnext"],"allowJs":true,"skipLibCheck":true,"strict":true,"noEmit":true,"esModuleInterop":true,"module":"esnext","moduleResolution":"bundler","resolveJsonModule":true,"isolatedModules":true,"jsx":"preserve","incremental":true,"plugins":[{"name":"next"}],"paths":{"@/*":["./src/*"]}},"include":["next-env.d.ts","**/*.ts","**/*.tsx"],"exclude":["node_modules"]}' > tsconfig.json
mkdir -p src/app
echo 'export default function RootLayout({ children }: { children: React.ReactNode }) { return (<html><body>{children}</body></html>) }' > src/app/layout.tsx
echo 'export default function Home() { return <main>Hello</main> }' > src/app/page.tsx
git add -A && git commit -m "init: Next.js scaffold"
git checkout -b dev
Record SMOKE_DIR for use in subsequent steps. The main session MUST cd to SMOKE_DIR before invoking /ship in Step 2.
IMPORTANT: Change working directory to $SMOKE_DIR before running /ship. All /ship operations (story creation, worktrees, merges) will target the temp repo, not ~/.claude.
cd "$SMOKE_DIR"
Use this hardcoded test fixture prompt:
Build a recipe search app. Next.js App Router, single page. Search bar at top that queries TheMealDB free API. Results show as cards with thumbnail, name, and category. Clicking a card opens a detail modal with full ingredients list and instructions. Users can bookmark recipes and see their bookmarks in a sidebar toggle.
Run:
/ship --quick "<the prompt above>"
The --quick flag skips critique/verify since the smoke test has its own validation.
Wait for /ship to complete before proceeding. Then cd back to the orchestration project root:
cd ~/.claude
After /ship completes, the generated code is on the dev branch of the temp repo. No cloning needed.
# Checkout dev branch which has the merged code
git -C "$SMOKE_DIR" checkout dev
APP_DIR="$SMOKE_DIR"
Verify the app has source files:
ls "$APP_DIR/src/" 2>/dev/null
If no source files exist, report FAIL and skip to Step 7 (teardown).
VALIDATOR_JSON=$(bash ~/.claude/scripts/project-validator.sh --project-root "$APP_DIR")
Parse the JSON result. Record:
validator_status: the top-level status fieldframework: the detected frameworkWIRING_JSON=$(bash ~/.claude/scripts/wiring-checklist.sh --project-root "$APP_DIR")
Parse the JSON result. Record:
wiring_status: the top-level status fieldwiring_score: the score field (e.g. "7/8")PASS if both conditions are met:
validator_status is "pass" (install + build + typecheck all succeed)FAIL otherwise.
Runs unconditionally, even if earlier steps failed.
Remove temp directory:
rm -rf "$SMOKE_DIR"
Archive the smoke-test epic (if one was created by /ship):
/cleanup epic-<id>
Print a summary table:
Smoke test: quality validation
Layer Status Output (last line)
--------------------------------------------------
install PASS/FAIL <last line of output>
build PASS/FAIL <last line of output>
typecheck PASS/FAIL <last line of output>
Wiring Check Status Matches
--------------------------------------------------
event_binding PASS/FAIL N
fetch_to_state PASS/FAIL N
click_to_state PASS/FAIL N
conditional_render PASS/FAIL N
shared_state PASS/FAIL N
toggle_pattern PASS/FAIL N
modal_close PASS/FAIL N
key_handler PASS/FAIL N
Wiring score: N/8
Build status: PASS/FAIL
Overall: PASS / FAIL
Legacy structural pipeline validation. Creates throwaway stories, runs planning agents, validates DB state and plan file structure, then tears down.
Load ToolSearch: select:mcp__gemini__pm_create_epic.
Call pm_create_epic(title="[SMOKE TEST] Pipeline validation").
Record the returned epic_id. The [SMOKE TEST] prefix makes it visually distinct in /roadmap output and identifiable if teardown fails.
Load ToolSearch: select:mcp__gemini__pm_create_story.
Create 3 stories under the smoke-test epic. Each targets real existing files that plan-writers can read but that will not actually be modified:
Story A — quick-fixer path:
pm_create_story(
title="[SMOKE] Add header comment to smoke-test skill",
epic_id=<epic_id>,
write_files=["skills/smoke-test/SKILL.md"],
agent="quick-fixer",
tasks=["Add a comment block to the top of skills/smoke-test/SKILL.md describing its purpose"]
)
Story B — architect path:
pm_create_story(
title="[SMOKE] Add validation notes to critique checklist",
epic_id=<epic_id>,
write_files=["refs/orch-critique-checklist.md"],
agent="architect",
tasks=[
"Add a 'Validation notes' section to refs/orch-critique-checklist.md",
"Include a cross-reference to the smoke-test skill"
]
)
Story C — multi-file architect path:
pm_create_story(
title="[SMOKE] Cross-reference smoke-test in cleanup skill",
epic_id=<epic_id>,
write_files=["skills/smoke-test/SKILL.md", "skills/cleanup/SKILL.md"],
agent="architect",
tasks=[
"Add a note in skills/cleanup/SKILL.md referencing the smoke-test teardown pattern",
"Add a corresponding back-reference in skills/smoke-test/SKILL.md"
]
)
Record all returned story IDs.
Load ToolSearch: select:mcp__gemini__pm_plan_stories.
Call pm_plan_stories(epic_id=<epic_id>, project_root="~/.claude").
This mirrors what /ship Step 1 does via the planner agent. Wait for it to complete.
Replicate the /ship Step 3b plan-writer agent pattern exactly.
refs/orch-critique-checklist.md — keep its full content for the agent prompts.pm_get_story(story_id) — read the detail file for tasks and write_files.plans/*.md once to get existing names. For each story, generate a unique plan file name with a smoke- prefix: plans/smoke-alpha.md, plans/smoke-beta.md, plans/smoke-gamma.md (or similar — avoid collisions with existing files).Launch one general-purpose background agent per story with run_in_background: true. Use this prompt template for each:
You are writing a plan file for story <story_id>: "<title>"
Agent: <agent>
Tasks: <task list from pm_get_story>
Write files: <write_files list>
Read files: <read_files list>
Output file: plans/<smoke-name>.md
## Critique Checklist
<full checklist content from refs/orch-critique-checklist.md>
## Instructions
1. Read the story's write_files to understand what exists today.
2. Read files referenced by tasks but not in write_files — these become read-only context.
3. Apply the critique checklist against the Gemini-planned tasks:
- If SIGNIFICANT issues found (missing files, scope creep, convention violations):
Return: "NEED_DECISION: <issue>\nOption A: <fix>\nOption B: <fix>"
- If MINOR gaps (edge cases, existing utilities): incorporate silently.
4. Write the plan file to plans/<smoke-name>.md with this structure:
# <story title>
Story: <story_id>
Agent: <agent>
## Context
<Brief description of what this story accomplishes>
## What changes
| File | Change |
|---|---|
| <write_file> | <description from tasks> |
## Read-only context
These files inform the implementation but should not be modified:
- `path/to/file` — why it's relevant
## Tasks
1. <task 1>
2. <task 2>
## Acceptance criteria
These define correctness independently of the implementation. Tests should verify these:
- <observable behavior 1>
- <observable behavior 2>
## Verification
- <how to verify the changes work>
5. Return: "DONE: plans/<smoke-name>.md"
Wait for all background agents to complete. For each result:
DONE: plans/<name>.md — Load ToolSearch: select:mcp__gemini__pm_update_story, then call pm_update_story(story_id=<id>, plan_file="plans/<name>.md").NEED_DECISION: <issue> — Log as a warning (do not block the smoke test for this). Note: "Plan-writer raised a decision — this is expected behavior for the pipeline but unexpected for a smoke test with trivial tasks."BLOCKED: <reason> — Log as a failure for this story.Run all checks independently. Collect pass/fail for each. Do NOT stop on failure — run every check regardless.
Initialize a results tracker:
checks = {
"plan_files_exist": { passed: 0, total: 0, details: [] },
"required_sections": { passed: 0, total: 0, details: [] },
"db_plan_file_set": { passed: 0, total: 0, details: [] },
"story_states": { states: [] },
"no_duplicate_tasks": { passed: 0, total: 0, details: [] },
"plan_file_nontrivial": { passed: 0, total: 0, details: [] }
}
For each story, read the plan file at its expected path. If the file exists and is readable, PASS. If not, FAIL with the missing path.
Each plan file must contain all of these headings:
## What changes (or ## Context)## Tasks## Acceptance criteriaRead each plan file and search for these headings. A plan file passes if all three are present.
Load ToolSearch: select:mcp__gemini__pm_get_story.
For each story, call pm_get_story(story_id) and read the detail file. Verify plan_file is non-null and matches the expected plans/smoke-*.md path.
For each story (from the same pm_get_story calls in 5c), record the current state. This is informational — note the actual state but do not mark it as a hard failure.
From each story's detail file, extract all task titles. Confirm no title appears more than once within a single story.
Each plan file should be >100 bytes. Read each file and check its size. Catches empty or stub files.
Clean up all smoke-test artifacts. This step runs unconditionally — even if validation checks failed.
Delete generated plan files:
rm plans/smoke-*.md 2>/dev/null || true
Load ToolSearch: select:mcp__gemini__pm_update_story,mcp__gemini__pm_update_epic.
For each story:
pm_update_story(story_id, state="done", force=True)
Close the epic:
pm_update_epic(epic_id, state="done")
Do NOT attempt to delete the epic from the DB (no such API). Closing the epic + marking stories done is sufficient.
Print a summary table with individual check results:
Smoke test: ship pipeline
Check Result
---------------------------------------
Plan files exist PASS/FAIL (N/M)
Required sections PASS/FAIL (details per file)
DB plan_file set PASS/FAIL (N/M)
Story states INFO: [list of states]
No duplicate tasks PASS/FAIL (N/M)
Plan file non-trivial PASS/FAIL (N/M)
Teardown: complete
Result: PASS / FAIL (N of M checks passed)
Count a check as PASS only if all sub-items within it passed. The overall result is PASS only if all non-INFO checks passed.
Generate interactive HTML diagrams (flowcharts, sequence diagrams, class diagrams, state diagrams, ER diagrams, Gantt charts, etc.) using Mermaid.js. Opens the result in the browser. Use when the user says "/diagram", "draw a flowchart", "make a diagram", "show me a flow", "sequence diagram", "class diagram", "state diagram", or any request for a visual diagram.
Graph-node skill: reads .ship-manifest.json (or story/epic IDs), launches one background agent per story to write plans/*.md files, applies critique checklist, updates DB with plan_file paths. Artifact-chain-aware version of /draft-plan. Consumes manifest from /plan-stories, produces plans/*.md for /env-preflight.
Spec-driven feature decomposition: accepts a FeatureSpec JSON (product, pattern, entity, fields, permissions, integrations, UI) and decomposes it into dependency-ordered stages using deterministic pattern DAGs. Creates stories via PM tools and writes .ship-manifest.json for downstream /draft-plans and /run-stories consumption. Use when the user says "/factory spec.json", "/factory --pattern crud-ui <inline JSON>", or "/factory --dry-run spec.json".
Merge a story worktree branch into its dev branch, push, clean up the worktree and branches, and mark the story done. Use when the user says "/merge-worktree", "/merge-worktree story-NNN", or "merge worktree".
Execute stories in parallel (where safe) using background agents, each working in an isolated git worktree branched from a shared dev branch. Handles dependency ordering, write-file conflict detection, and reports results. Use when the user says "/run-stories", "/run-stories story-NNN", "/run-stories epic-NNN", or any combination of story and epic IDs.
Bridge between presearch briefings (or inline args) and the /draft-plans skill. Parses inputs, dispatches the planner agent to create an epic and stories in the DB, and writes .ship-manifest.json for downstream consumption. Three invocation modes: briefing path, inline title+items, or resume an existing epic. Also handles execute mode for existing plan files (bypasses planner, creates DB entries directly). Use when the user says "/plan-stories presearch/topic.md", "/plan-stories \"Title\" 1. Feature 2. Feature", or "/plan-stories epic-NNN".