nanostack

分类: 工具与效率 | 上传者: garagon | 下载: 0 | 版本: v1.0（最新）

当用户询问可用的工作流程技能、想要了解工程工作流程或引用“nanostack”时使用。也在 /nanostack 上触发。

更新日志: Source: GitHub https://github.com/garagon/nanostack

目录结构

当前层级: tree/main/

📁 .claude-plugin/
- 📄 plugin.json 518 B
📁 .cursor-plugin/
- 📄 plugin.json 273 B
📁 .github/
- 📁 ISSUE_TEMPLATE/
  - 📄 bug_report.yml 1.3 KB
  - 📄 config.yml 230 B
  - 📄 feature_request.yml 1.2 KB
- 📁 workflows/
  - 📄 e2e.yml 9.7 KB
  - 📄 lint.yml 173.1 KB
- 📄 CODEOWNERS 42 B
- 📄 pull_request_template.md 332 B
📁 adapters/
- 📄 claude.json 498 B
- 📄 codex.json 557 B
- 📄 cursor.json 599 B
- 📄 gemini.json 581 B
- 📄 opencode.json 602 B
📁 bin/
- 📁 lib/
  - 📄 artifact-schemas.sh 6.7 KB
  - 📄 artifact-trust.sh 3.3 KB
  - 📄 audit.sh 1.3 KB
  - 📄 cache.sh 2.3 KB
  - 📄 git-context.sh 436 B
  - 📄 html-escape.sh 2.6 KB
  - 📄 phases.sh 17.8 KB
  - 📄 portable.sh 1.6 KB
  - 📄 preflight.sh 1.2 KB
  - 📄 pricing.sh 556 B
  - 📄 skill-finalize.sh 3.4 KB
  - 📄 skill-preamble.sh 3.3 KB
  - 📄 solutions-index.sh 4.4 KB
  - 📄 store-path.sh 589 B
  - 📄 telemetry.sh 21.8 KB
  - 📄 visual-render.sh 22.3 KB
- 📄 about.sh 4.3 KB
- 📄 analytics.sh 9.8 KB
- 📄 budget.sh 4.3 KB
- 📄 capture-failure.sh 1.6 KB
- 📄 capture-learning.sh 1.0 KB
- 📄 check-adapters.sh 15.0 KB
- 📄 check-custom-skill.sh 7.1 KB
- 📄 circuit.sh 3.9 KB
- 📄 create-skill.sh 5.2 KB
- 📄 detect-legacy-setup.sh 4.3 KB
- 📄 discard-sprint.sh 2.3 KB
- 📄 doctor.sh 9.6 KB
- 📄 find-artifact.sh 6.3 KB
- 📄 find-solution.sh 10.4 KB
- 📄 gather-subject.sh 5.9 KB
- 📄 graduate.sh 11.5 KB
- 📄 init-config.sh 2.0 KB
- 📄 init-project.sh 12.0 KB
- 📄 init-stack.sh 4.4 KB
- 📄 loop-guard.sh 3.0 KB
- 📄 nano-doctor.sh 26.4 KB
- 📄 next-step.sh 11.3 KB
- 📄 pattern-report.sh 8.9 KB
- 📄 render-artifact.sh 81.6 KB
- 📄 resolve.sh 29.3 KB
- 📄 restore-context.sh 4.5 KB
- 📄 save-artifact.sh 7.2 KB
- 📄 save-setup-artifact.sh 5.7 KB
- 📄 save-solution.sh 2.3 KB
- 📄 scope-drift.sh 2.0 KB
- 📄 session.sh 26.3 KB
- 📄 sprint-journal.sh 6.9 KB
- 📄 sprint-metrics.sh 3.7 KB
- 📄 telemetry-config.sh 5.5 KB
- 📄 telemetry-log.sh 7.7 KB
- 📄 token-analyzer.py 19.9 KB
- 📄 token-report.sh 1.5 KB
- 📄 upgrade.sh 4.6 KB
- 📄 validate-dependencies.sh 1.7 KB
📁 ci/
- 📄 check-custom-stack-examples.sh 13.2 KB
- 📄 check-examples.sh 7.4 KB
- 📄 check-visual-artifact-templates.sh 9.3 KB
- 📄 e2e-adapter-freshness.sh 19.0 KB
- 📄 e2e-artifact-trust.sh 12.6 KB
- 📄 e2e-custom-routing.sh 17.8 KB
- 📄 e2e-custom-stack-examples.sh 16.4 KB
- 📄 e2e-custom-stack-flows.sh 19.9 KB
- 📄 e2e-delivery-matrix.sh 11.9 KB
- 📄 e2e-examples.sh 13.7 KB
- 📄 e2e-graph-aware-session.sh 29.0 KB
- 📄 e2e-onboarding-flows.sh 16.2 KB
- 📄 e2e-structured-artifacts.sh 9.6 KB
- 📄 e2e-think-archetypes.sh 17.4 KB
- 📄 e2e-think-flows.sh 12.9 KB
- 📄 e2e-user-flows.sh 31.8 KB
- 📄 e2e-visual-artifacts.sh 91.3 KB
📁 commands/
- 📄 nano-update.md 419 B
📁 compound/
- 📄 SKILL.md 8.1 KB
📁 conductor/
- 📁 agents/
  - 📄 openai.yaml 334 B
- 📁 bin/
  - 📄 sprint.sh 23.5 KB
- 📄 SKILL.md 10.3 KB
📁 doctor/
- 📄 SKILL.md 6.4 KB
📁 examples/
- 📁 api-healthcheck/
  - 📄 README.md 6.6 KB
  - 📄 server.js 2.0 KB
- 📁 cli-notes/
  - 📄 notes.sh 1.6 KB
  - 📄 README.md 5.6 KB
- 📁 custom-skill-template/
  - 📁 audit-licenses/
    
    📁 agents/
    
    📄 openai.yaml 370 B
    
    📁 bin/
    
    📄 audit.sh 4.4 KB
    
    📄 smoke.sh 1.9 KB
    
    📄 SKILL.md 5.0 KB
  - 📄 README.md 5.6 KB
- 📁 custom-stack-template/
  - 📁 compliance-release/
    
    📁 skills/
    
    📁 license-audit/
    
    📁 agents/
    
    📄 openai.yaml 323 B
    
    📁 bin/
    
    📄 audit.sh 5.9 KB
    
    📄 smoke.sh 4.8 KB
    
    📄 SKILL.md 2.8 KB
    
    📁 privacy-check/
    
    📁 agents/
    
    📄 openai.yaml 325 B
    
    📁 bin/
    
    📄 check.sh 6.2 KB
    
    📄 smoke.sh 5.7 KB
    
    📄 SKILL.md 3.4 KB
    
    📁 release-readiness/
    
    📁 agents/
    
    📄 openai.yaml 332 B
    
    📁 bin/
    
    📄 smoke.sh 7.5 KB
    
    📄 summarize.sh 4.9 KB
    
    📄 SKILL.md 4.4 KB
    
    📄 README.md 9.6 KB
    
    📄 stack.json 1.8 KB
  - 📄 README.md 2.9 KB
- 📁 starter-todo/
  - 📄 index.html 2.5 KB
  - 📄 README.md 4.4 KB
- 📁 static-landing/
  - 📄 index.html 2.7 KB
  - 📄 README.md 6.1 KB
- 📄 README.md 2.9 KB
📁 feature/
- 📁 bin/
  - 📄 enforce-sprint.sh 2.6 KB
- 📄 SKILL.md 5.0 KB
📁 guard/
- 📁 agents/
  - 📄 openai.yaml 319 B
- 📁 bin/
  - 📄 budget-gate.sh 1.2 KB
  - 📄 check-dangerous.sh 11.4 KB
  - 📄 check-write.sh 11.1 KB
  - 📄 phase-gate.sh 9.1 KB
- 📄 rules.json 12.5 KB
- 📄 SKILL.md 6.4 KB
📁 help/
- 📄 SKILL.md 3.0 KB
📁 plan/
- 📁 agents/
  - 📄 openai.yaml 319 B
- 📁 references/
  - 📄 product-standards.md 3.3 KB
  - 📄 stack-defaults.md 5.5 KB
- 📁 templates/
  - 📄 plan-template.md 821 B
  - 📄 product-spec.md 941 B
  - 📄 technical-spec.md 1.2 KB
- 📄 SKILL.md 13.5 KB
📁 qa/
- 📁 agents/
  - 📄 openai.yaml 284 B
- 📁 bin/
  - 📄 screenshot.sh 1.1 KB
- 📄 SKILL.md 12.8 KB
📁 reference/
- 📄 agent-agnostic-delivery-spec.md 22.6 KB
- 📄 artifact-schema.md 15.3 KB
- 📄 conflict-precedents.md 4.0 KB
- 📄 custom-stack-contract.md 16.3 KB
- 📄 custom-stack-examples-technical-spec.md 20.7 KB
- 📄 diarization-schema.md 2.4 KB
- 📄 host-adapter-schema.md 5.5 KB
- 📄 plain-language-contract.md 5.1 KB
- 📄 session-state-contract.md 6.8 KB
- 📄 solution-schema.md 4.2 KB
- 📄 v1-delivery-experience-technical-spec.md 16.3 KB
- 📄 visual-artifact-contract.md 12.8 KB
📁 review/
- 📁 agents/
  - 📄 openai.yaml 302 B
- 📁 bin/
  - 📄 suggest-security.sh 1.1 KB
- 📄 checklist.md 2.0 KB
- 📄 SKILL.md 13.1 KB
📁 security/
- 📁 agents/
  - 📄 openai.yaml 330 B
- 📁 references/
  - 📄 owasp-checklist.md 5.0 KB
- 📁 templates/
  - 📄 security-report.md 1.8 KB
- 📄 SKILL.md 13.9 KB
📁 ship/
- 📁 agents/
  - 📄 openai.yaml 289 B
- 📁 bin/
  - 📄 pre-ship-check.sh 1.9 KB
  - 📄 quality-check.sh 2.9 KB
- 📁 references/
  - 📄 repo-quality-standards.md 1.0 KB
- 📁 templates/
  - 📄 pr-template.md 379 B
- 📄 SKILL.md 16.3 KB
📁 skills/
- 📄 conductor 12 B
- 📄 guard 8 B
- 📄 plan 7 B
- 📄 qa 5 B
- 📄 review 9 B
- 📄 security 11 B
- 📄 ship 7 B
- 📄 think 8 B
📁 start/
- 📁 references/
  - 📄 onboarding-contract.md 7.8 KB
- 📄 SKILL.md 11.6 KB
📁 telemetry-worker/
- 📁 migrations/
  - 📄 0001_init.sql 1.2 KB
  - 📄 0002_add_observational_fired.sql 550 B
- 📁 src/
  - 📄 index.ts 10.3 KB
- 📄 bun.lock 21.5 KB
- 📄 package.json 471 B
- 📄 README.md 3.7 KB
- 📄 tsconfig.json 447 B
- 📄 verify-security.sh 7.9 KB
- 📄 wrangler.toml 1.0 KB
📁 think/
- 📁 agents/
  - 📄 openai.yaml 311 B
- 📁 presets/
  - 📄 default.md 1.4 KB
  - 📄 design.md 3.0 KB
  - 📄 devex.md 3.4 KB
  - 📄 eng.md 3.5 KB
  - 📄 garry.md 4.6 KB
  - 📄 yc.md 2.7 KB
- 📁 references/
  - 📄 archetypes.md 16.4 KB
  - 📄 brief-template.md 1.4 KB
  - 📄 cognitive-patterns.md 5.9 KB
  - 📄 forcing-questions.md 5.5 KB
  - 📄 latent-vs-deterministic.md 5.9 KB
  - 📄 observational-patterns.md 5.2 KB
  - 📄 search-before-building.md 6.0 KB
- 📄 SKILL.md 40.3 KB
📄 .gitignore 529 B
📄 AGENTS.md 4.1 KB
📄 CODE_OF_CONDUCT.md 2.4 KB
📄 CONTRIBUTING.md 2.9 KB
📄 EXTENDING.md 13.2 KB
📄 gemini-extension.json 172 B
📄 GEMINI.md 22 B
📄 LICENSE 10.5 KB
📄 llms.txt 4.5 KB
📄 README.es.md 19.6 KB
📄 README.md 52.1 KB
📄 SECURITY.md 5.7 KB
📄 setup 22.4 KB
📄 SKILL.md 7.9 KB
📄 TELEMETRY.md 19.6 KB
📄 TROUBLESHOOTING.es.md 6.6 KB
📄 TROUBLESHOOTING.md 10.4 KB
📄 VERSION 6 B
📄 ZEN.md 680 B

SKILL.md

---
name: nanostack
description: Use when the user asks about available workflow skills, wants an overview of the engineering workflow, or references "nanostack". Also triggers on /nanostack.
---

# Nanostack — Engineering Workflow Skills

You have access to a set of composable engineering workflow skills. Each skill is a folder with supporting files — read them as needed for context.

## Available Skills

| Skill | When to use | Modes | Key files |
|-------|-------------|-------|-----------|
| `/think` | Before planning — strategic product thinking, premise validation, scope decisions. | — | `think/references/forcing-questions.md`, `think/references/cognitive-patterns.md` |
| `/nano` | Before starting any non-trivial work. Produces a scoped, actionable plan. | — | `plan/templates/plan-template.md` |
| `/review` | After code is written. Two-pass review + scope drift detection + conflict resolution. | `--quick` `--standard` `--thorough` | `review/checklist.md`, `reference/conflict-precedents.md` |
| `/qa` | To verify code works. Browser-based testing with Playwright, plus root-cause debugging. | `--quick` `--standard` `--thorough` | `qa/bin/screenshot.sh` |
| `/security` | Before shipping. OWASP Top 10 + STRIDE + variant analysis + conflict detection. | `--quick` `--standard` `--thorough` | `security/references/owasp-checklist.md`, `security/templates/security-report.md` |
| `/ship` | To create PRs, merge, deploy, and verify. Generates sprint journal on success. | — | `ship/templates/pr-template.md` |
| `/guard` | When working near production, destructive operations, or sensitive systems. | — | `guard/bin/check-dangerous.sh` |
| `/feature` | Add a feature to an existing project. Skips /think, goes straight to plan → build → review → security → qa → ship. | — | `feature/SKILL.md` |
| `/conductor` | Orchestrate parallel agent sessions through a sprint. Coordinate task claiming and artifact handoff. | `start` `claim` `complete` `status` | `conductor/bin/sprint.sh` |
| `/nano-run` | First-time setup. Configures stack, permissions, and preferences conversationally. Guides first sprint. | — | `start/SKILL.md` |
| `/nano-help` | Quick reference for all nanostack commands and how to use them. | — | `help/SKILL.md` |

## Workflow Order

The default workflow is: `/think` → `/nano` → build → `/review` → `/qa` → `/security` → `/ship`

With `/conductor`, review + qa + security run **in parallel** — they all depend on build, not on each other:

```
think → plan → build ─┬─ review  ─┐
                      ├─ qa       ├─ ship
                      └─ security ─┘
```

Activate `/guard` at any point when operating near production or sensitive systems.

## Zen

Read `ZEN.md` for the full set of principles. When in doubt about a decision during any skill, consult it. The short version:

- Question the requirement before writing the code.
- Delete what shouldn't exist. Don't optimize what's left until you do.
- Narrow the scope, not the ambition.
- Fix it or ask. Never ignore it.
- Security is not a tradeoff. It is a constraint.
- The output should look better than what was asked for.

## Intensity Modes

Skills `/review`, `/security`, and `/qa` support intensity modes:

| Mode | Flag | When | Confidence |
|------|------|------|-----------|
| **Quick** | `--quick` | Trivial changes (typos, config, docs) | 9/10 — only the obvious |
| **Standard** | (default) | Normal changes | 7/10 — anything reasonable |
| **Thorough** | `--thorough` | Critical changes (auth, payments, infra) | 3/10 — flag everything suspicious |

Skills auto-suggest a mode based on the diff, but the user always decides.

## Artifact Persistence

**Saving artifacts is not optional.** Every skill must save its artifact after completing.

Skills automatically save their output to `.nanostack/` after every run:

```bash
.nanostack/<phase>/<timestamp>.json
```

This enables:
- **Scope drift detection** — `/review` compares planned vs actual files
- **Conflict detection** — `/review` and `/security` cross-reference each other's findings
- **Sprint journals** — `/ship` generates a journal entry from all phase artifacts
- **Trend tracking** — Are security findings decreasing over time?

Auto-saving is on by default. The user can disable it by setting `auto_save: false` in `.nanostack/config.json`.

Artifacts are validated before saving: `save-artifact.sh` rejects invalid JSON, missing required fields (`phase`, `summary`), and phase mismatches.

To discard artifacts from a bad session: `bin/discard-sprint.sh` (removes artifacts and journal entry for the current project and date).

## Conflict Resolution

When skills produce contradictory guidance (e.g., `/review` says "more error detail" but `/security` says "minimize error exposure"), the conflict resolution framework applies:

1. **Security has default precedence** — unless the risk is theoretical or internal-only
2. **Context determines final precedence** — public-facing app vs internal tool vs startup vs compliance
3. **Conflicts are documented, not silenced** — every resolution is recorded in the skill artifact

Read `reference/conflict-precedents.md` for known conflict patterns and pre-defined resolutions.

## Project Config

On first use in a project, run `bin/init-config.sh --interactive` to create `.nanostack/config.json`. This stores:

- **Installed agents** — auto-detected (claude, codex, cursor, opencode, gemini)
- **Detected stack** — node, go, python, docker
- **Preferences** — default intensity mode, auto-save, conflict precedence

If config exists, read it at the start of any skill to adapt behavior:
```bash
bin/init-config.sh  # outputs current config or {} if none
```

Skills use config for:
- `/review`, `/qa`, `/security`: read `preferences.default_intensity` instead of always defaulting to standard
- `/security`: read `preferences.conflict_precedence` to determine who wins in cross-skill conflicts
- `/security`: read `detected` to skip irrelevant checks (don't scan for Python vulns in a Go project)

Per-skill configs (`security/config.json`, `guard/config.json`) store skill-specific settings and are read by that skill only.

## Universal Rules

- **Read before acting.** Every skill must read the relevant code/diff/config before producing output. Never analyze blind.
- **Boil the lake, not the ocean.** When completeness costs minutes more than shortcuts, do the complete thing. When it costs days, don't.
- **Log failures, not just successes.** When something goes wrong during a skill (script error, wrong approach, unexpected project behavior), capture it:
  ```bash
  ~/.claude/skills/nanostack/bin/capture-failure.sh <skill> "<what went wrong>" "<what was tried>" "<what fixed it>"
  ```
  Failures compound into knowledge. Next sprint, the same mistake is avoided. This does not require /compound or a successful ship — just log and move on.

## Proactive Triggers

Suggest skills when context matches — don't wait for the user to remember:

| Trigger | Suggest |
|---------|---------|
| User says "what should I build" / unclear on direction | `/think` |
| Task touches 3+ files or user says "how should I approach this" | `/nano` |
| User says "done", "finished", "ready for review" | `/review` |
| User says "does this work", "test this", bug report | `/qa` |
| Pre-ship, user says "ready to deploy", or diff touches auth/env/infra | `/security` |
| User says "create PR", "merge", "ship it" | `/ship` |
| Destructive commands, production access, or sensitive operations detected | `/guard` |

## Usage Rules

- Start with `/think` for new products or when the "what" is unclear
- Run `/nano` before building anything that touches more than 3 files
- Run `/review` on your own code — the adversarial pass catches what you missed
- `/security` is not optional before shipping to production
- `/guard` is on-demand — activate it, don't leave it always on
- Skills compose: `/qa` can invoke `/security` checks, `/ship` can invoke `/review`

登录后下载/点赞/收藏 ❤ 194 | ★ 0

nanostack

目录结构

SKILL.md

举报内容

提示