app-idea
Run a new or existing consumer/AI app idea through the golden-track playbook (~/wiki/playbook/apps.md) from intake to verdict, write a verdict card, stop for the owner, and ratchet owner overrides back into the doctrine. Use when the user shares an app idea, asks "should I build X", "audit <app>", "re-score <app>", or pastes an idea in a product context. Never builds; never promotes a rule on its own.
- License 许可
- MIT
- 所属仓库
- Necmttn/skills
- 最近更新
- 2026-11-07
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- app-idea
- description
- Run a new or existing consumer/AI app idea through the golden-track playbook (~/wiki/playbook/apps.md) from intake to verdict, write a verdict card, stop for the owner, and ratchet owner overrides back into the doctrine. Use when the user shares an app idea, asks "should I build X", "audit <app>", "re-score <app>", or pastes an idea in a product context. Never builds; never promotes a rule on its own.
- user-invocable
- true
02 资源文件 7
- 脚本scripts/check-playbook.sh2.6 KB
- 参考文档templates/verdict-card.md1.3 KB
- 参考文档tests/fixtures/bad/playbook/apps.md769 B
- 参考文档tests/fixtures/bad/playbook/verdicts/2026-01-01-sample.md522 B
- 参考文档tests/fixtures/good/playbook/apps.md773 B
- 参考文档tests/fixtures/good/playbook/verdicts/2026-01-01-sample.md551 B
- 脚本tests/run.sh741 B
03 SKILL.md 正文
/app-idea
Intake -> verdict for one app idea. The rules live in ~/wiki/playbook/apps.md; this file holds the procedure only. If a rule you need is not on that page, apply it as candidate in the card and say so. Do not add rules here.
0. Guardrails
- Write under
~/wiki/playbook/**and append to~/wiki/log.mdvia shell (cat > file <<'EOF') withALLOW_MAIN_WRITE=1exported; the write-guard hook blocks Write/Edit under~/wiki. Nevergit commitin~/wiki. - One card per run. Never overwrite an existing card. A re-score gets a new dated card that links the old one.
- Statuses move only on an explicit owner instruction (step 7). Dry runs (
run: dry-run) never move a status. - Never edit
~/wiki/raw/**or~/Projects/apps/docs/playbooks/*.
1. Orient (always)
Read, in order: ~/wiki/playbook/apps.md, ~/wiki/index.md, ~/wiki/queries/2026-08-28-app-ideas-from-knowledge-base.md, the last 20 lines of ~/wiki/log.md.
Existing app: also read ~/Projects/apps/apps/<app>/docs/METRICS.md, ~/Projects/apps/docs/experiments/<app>/, ~/Projects/apps/docs/reports/<app>/ when they exist, and the wiki entity page if one exists.
2. Intake (gate I-01)
Restate the idea in one sentence a stranger understands. Cannot? Stop. Ask the owner for the sentence. Do not proceed on a paragraph.
Decide kind: new (no shipped app) or existing (an app dir under ~/Projects/apps/apps/ or a wiki entity page exists).
3. Evidence sweep
new: WebSearch for competitor revenue, App Store rank, reviews, paid ads, creator content. If~/.cache/appllama/exists, read cached app walks before spending credits (see~/Projects/apps/docs/playbooks/appllama-research.md). Every receipt is a URL.existing: repo artifacts and wiki entity pages only. Every receipt is a file path. No web.- No receipts found: the verdict is
parkand the card names the single receipt to go get.
4. Score
Apply every I- rule. Fill the four paying-intent bands with an estimate and a one-phrase basis. Score evidence / monetization / built-in loop 0-5. Name the hardest risk and one numeric, dated kill criterion. Cite each rule you leaned on by id with its current status.
5. Card
Copy templates/verdict-card.md (relative to this skill dir) to ~/wiki/playbook/verdicts/YYYY-MM-DD-<slug>.md. Fill every {{...}}. Set run: to normal|dry-run|audit. Then lint:
<this-skill-dir>/scripts/check-playbook.sh ~/wiki
It must print OK. Fix before presenting.
Append to ~/wiki/log.md: ## [DATE] app-idea | <slug> | <verdict> plus one line naming the card path and the rule ids applied.
6. Stop
Show the card in chat: one-liner, verdict, scores, kill criterion, rules applied. End the turn. Build starts only on an explicit go, via ~/Projects/apps/docs/playbooks/new-app.md.
7. Feedback loop (owner-triggered only)
On an owner override (an edited ## Owner override block, or a chat instruction):
- Score or verdict change: update the card's
updated:and the field; record the override in the block. - Rule promotion ("that rule is right", "make P-02 decided"): append
- DATE | P-02 | candidate -> decided | <owner reason> | <card path>under## Decision loginapps.md, flip the rule line's status and date, bumpupdated:. - Verified ("we measured it"): require a path to an experiment file with a DECIDED line or a weekly report; flip to
verifiedwith that path as evidence. - Retire: move the rule line under
## Retired, strike the text with~~, addReason:; log line... -> retired. - Re-lint. Append a
log.mdline.
Dry-run protocol
When asked to dry-run: set run: dry-run, complete steps 1-6, skip step 7 entirely.
04 同仓库更多 Skills
wiki
Recall from the personal LLM wiki at ~/wiki. Use when the user asks what the wiki says, wants prior thinking/intuition on a concept, or when grounding a design or decision in previously synthesized knowledge. Read-only - never writes to the wiki.
wiki-ingest
Curate a source into the personal LLM wiki at ~/wiki following its SCHEMA.md - a file, URL, research output, ax session, or a pending ~/wiki/inbox/ brief. Use when the user says "add this to the wiki", "ingest this", "promote the inbox", or after research worth keeping long-term.
wiki-harvest
Propose-only sweep of recent coding sessions (via the ax CLI) for recurring concepts that are missing from, or should update, the personal LLM wiki at ~/wiki. Writes candidate briefs to ~/wiki/inbox/ only - never wiki pages. Use on schedule or when the user asks "harvest the wiki" or "what concepts am I repeating".
no-ai-slop
Strip AI-slop patterns from a draft while preserving the writer's voice, or detect slop without rewriting. Use as the mandatory final pass on any copy before shipping, when the user wants a draft clearer/less AI-sounding, or when asked "is this AI slop?", "audit this writing", "does this read as AI".
aso-appstore-screenshots
Generate high-converting App Store screenshots by analyzing your app's codebase, discovering core benefits, and creating ASO-optimized screenshot images using Nano Banana Pro.
wrap-up
Close down a work session or agent pane cleanly - review the work against every dimension (complete, edge cases, correct, consistent, production ready, goals met), land or file everything of value (follow-up tickets for loose ends, concerns, debt), add human-verification items to a `uat`-labeled checklist issue (builds to verify, flows to smoke), then do safe-only housekeeping (pushed worktrees, own background processes, created sims) and report what was closed, filed, and kept. Use when a session's work is done and the user says "wrap up", "close this down", "can we close this", "clean up and land it", when a fleet run or chunk finishes, or before ending a long session that minted resources.