AGENT SKILL

simplified-architect

Use only when the user explicitly says "simple architect".

License 许可
Apache-2.0
最近更新
2026-11-07
Stars
★ 2
Forks
0

01 前言 Front Matter

name
simplified-architect
description
Use only when the user explicitly says "simple architect".

02 资源文件 1

03 SKILL.md 正文

Simplified Architect

Create a PM-level spec and a phased implementation plan directly through two deliberate rounds of user questioning.

Shared guidance

  • Spec rules: ../../references/spec-guidance.md
  • Plan rules: ../../references/plan-guidance.md

Read both files before drafting or revising artifacts.

Paths

  • Output directory: ./output/architect/
  • {PlanName} is PascalCase derived from the task.
  • Spec: ./output/architect/{PlanName}-spec.md
  • Plan: ./output/architect/{PlanName}-plan.md

Core planning principle

Prefer the simplest clean design that satisfies the confirmed goal.

When broader scenario coverage, defensive flexibility, or an edge case would make the design harder to understand or implement:

  1. Treat it as out of scope in the default design.
  2. Record and explain the resulting gap plainly.
  3. Ask whether the user wants that gap promoted into scope.
  4. Add the complexity only when the user chooses it or the confirmed goal cannot work safely without it.

Do not silently ignore correctness, security, data-loss, or compatibility requirements. Simplicity means avoiding speculative complexity, not accepting a broken design.

Mode detection

  • Create: no matching spec and plan exist. Run the full workflow.
  • Revise: matching artifacts exist. Read them, interview the user about the requested change, and revise both when product intent changes.

Workflow

1. Scan the workspace

Inspect the relevant repository structure, code, tests, configuration, documentation, and existing patterns. Learn technical facts from the workspace rather than asking the user to supply them.

2. First questioning round: understand the purpose

Ask one question at a time through the host picker. Prefer choices when they capture the real decision; use free text when they do not.

Continue until all material areas are clear:

  • Goal and problem being solved
  • Users and audience
  • User-visible behavior
  • Scope and non-goals
  • Success criteria
  • Constraints and integrations
  • Behavioral defaults
  • Compatibility or migration expectations
  • Failure behavior and meaningful risks
  • Simplicity trade-offs and intentionally uncovered edge cases

Questions should be thorough, grounded in the scan, and limited to decisions only the user can make. Do not impose a fixed question count or stop because a small quota was reached.

Before drafting, summarize the understood intent internally and check that no answer conflicts with another. Ask another question whenever the purpose or acceptance boundary is still ambiguous.

3. Draft the spec and plan

Write both artifacts using the shared guidance.

Use Pragmatic as the plan approach unless the user explicitly requests another label. Keep the plan focused on the simplest implementation that fulfills the spec. Put excluded scenarios in the spec's Non-Goals and any implementation consequence in plan risks or open questions.

4. Second questioning round: challenge the drafts

Read the complete draft spec and plan together. Look for:

  • Assumptions that became visible only after decomposition
  • Conflicts between product intent and implementation steps
  • Missing acceptance boundaries
  • Dependencies or migrations that change the scope
  • Edge cases whose omission creates a meaningful gap
  • Places where a simpler design is possible
  • Places where the simple path has a trade-off the user should consciously accept

Ask one focused question at a time about every material decision found. Always perform this round; if the drafts expose no unresolved decision, ask the user to confirm the summarized simple path and its explicitly listed gaps.

5. Revise both artifacts

Apply the second-round answers to both spec and plan wherever relevant. Re-run the shared pre-write checklists and fix contradictions, stale open questions, and scope drift.

After revision, proceed directly to the report without adding a review score.

6. Report

Report:

  • Spec and plan paths
  • Brief phase and step counts
  • The chosen simple approach
  • Edge cases or scenarios intentionally left out
  • Any unresolved question that the user explicitly deferred

Communication

  • Ask every user question through the host picker.
  • Ask exactly one question per picker call.
  • Keep progress updates brief.

04 同仓库更多 Skills

LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

my-skill

SKILL_DESCRIPTION — when should Claude load this skill?

★ 2
LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

my-skill

SKILL_DESCRIPTION — when should Claude load this skill?

★ 2
LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

humanizer

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.

★ 2
LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

deep-research

Use when the user needs multi-source research with citation tracking, evidence persistence, and structured report generation. Triggers on "deep research", "comprehensive analysis", "research report", "compare X vs Y", "analyze trends", or "state of the art". Not for simple lookups, debugging, or questions answerable with 1-2 searches.

★ 2
LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

ai-local-diffusion-invoker

Generate images locally on the user's own NVIDIA GPU by invoking a separately-installed LocalAIExecution `localai` CLI (FLUX text-to-image) through its stable --json subprocess contract. This skill vendors NONE of LocalAIExecution's code — it discovers the external tool at runtime and shells out to it. Use when the user wants to create/generate an image, render a picture from a text prompt, run local text-to-image / diffusion / FLUX, or make art on their own GPU without cloud or paid APIs. Triggers on: 'generate an image of X', 'create a picture of X', 'render X locally', 'make art of X', 'text-to-image X', 'run FLUX on X', 'local diffusion X'.

★ 2
LiorZivi
LiorZivi/ZiviDevelopmentMarketplace

telegram-remote

Activate, operate, and tear down a bidirectional bridge between this Copilot CLI session and a Telegram bot DM so the user can step away from the terminal. Works from any repository and needs only a Telegram bot token plus chat ID. Posts meaningful progress, routes questions, uploads requested local files, and injects Telegram replies back into the session. Invoke with no args to activate or with 'end' to close an active session. Activation triggers: 'telegram-remote', 'i'm stepping away', 'ping me on telegram', 'post progress to telegram', 'afk mode', 'continue on telegram', 'headed into a meeting'. End triggers: 'telegram-remote end', 'end-telegram-remote', 'stop-telegram-remote', 'i'm back at the terminal', 'back from afk', 'stop posting to telegram', 'disable remote mode'.

★ 2