# Using Deconstruct Toolkit

> |

- **Type:** Skill
- **Install:** `agentstack add skill-kouko-monkey-skills-using-deconstruct-toolkit`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [kouko](https://agentstack.voostack.com/s/kouko)
- **Installs:** 0
- **Category:** [Agent Skills](https://agentstack.voostack.com/c/agent-skills)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [kouko](https://github.com/kouko)
- **Source:** https://github.com/kouko/monkey-skills/tree/main/deconstruct-toolkit/skills/using-deconstruct-toolkit

## Install

```sh
agentstack add skill-kouko-monkey-skills-using-deconstruct-toolkit
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Using Deconstruct Toolkit

Help the user find the right deconstruction skill for their artifact.
Detect what they brought, what they want surfaced, then dispatch to
the best-fit sibling skill with the appropriate lens combination
preselected.

## Boundary check first

Before routing, run a 3-question boundary check. If any answer is
"yes," do NOT route into this toolkit — redirect.

| Question | If yes, redirect to |
|---|---|
| Is the target source code, a code repository, or a build artifact? | `sourceatlas` (impact / flow / overview / pattern / deps) |
| Does the user want help thinking through their *own* problem (not analyzing an external artifact)? | `philosophers-toolkit` |
| Does the user want to **produce** copy / docs / design / a slide deck (forward direction, not reverse)? | `copywriting-toolkit` / `docs-team` / `design-team` / `slides-toolkit` |

If all three are "no," proceed to routing.

## Routing Guide

Ask the user ONE question: **"What did you bring, and what do you want to surface?"**

Then match the answer through the two-axis lookup below.

### Axis 1 — Artifact type

| Artifact | Default skill | Default lens combination |
|---|---|---|
| Marketing copy / landing page / advertising | `artifact-deconstruct` | lens-persuasion + lens-rhetoric |
| Document pack / playbook / SOP / onboarding kit | `artifact-deconstruct` | lens-genre + 6-dimension full |
| UI / app onboarding flow / website screen | `artifact-deconstruct` | lens-ux + lens-persuasion |
| Long-form argument / op-ed / proposal / political text | `argument-deconstruct` | Toulmin + Burke pentad + warrant surface |
| Strategy memo / policy brief / social-media thread (suspect hidden assumptions) | `assumption-surface` | reverse-Toulmin + symptomatic reading |
| Speech / political address | `artifact-deconstruct` | lens-rhetoric + lens-frame |
| Literature / film / advertising imagery | `artifact-deconstruct` | lens-semiotic + lens-frame |
| Slide deck / presentation | `artifact-deconstruct` | lens-rhetoric + lens-genre |

### Axis 2 — User intent override

The artifact type sets the default; the user's stated intent can override which skill to dispatch:

| User says | Override to |
|---|---|
| "拆解設計" / "deconstruct the design" / "為什麼這份寫得這麼好" | `artifact-deconstruct` (full 6-lens × 6-dim treatment) |
| "找隱性假設" / "surface the hidden assumptions" / "what is this *assuming*" / "stress-test the claims" | `assumption-surface` (atomic, fast) |
| "拆解論證" / "find the warrant" / "is this argument valid" / "where does this argument fail" | `argument-deconstruct` (Toulmin focus) |
| "找說服機制" / "what's manipulating me here" / "spot the dark patterns" | `artifact-deconstruct` with lens-persuasion + lens-ux preselected |

If artifact type and user intent disagree, **user intent wins**.

## Three filters before dispatching

After tentative routing, run these filters:

1. **Length filter** — if the artifact is  Dispatching to `artifact-deconstruct` with `lens-persuasion + lens-rhetoric` preselected, language=Japanese, register=consumer-marketing → variants `-ja`. Running now.

## Rules

- Do NOT perform deconstruction in this skill — only route
- Do NOT explain lens content in detail — let the dispatched skill do that
- Recommend ONE skill, not multiple. If the user wants two passes (e.g., assumption-surface first, then artifact-deconstruct), do them sequentially
- If the user already specified the skill, skip routing and invoke directly
- If none of the skills fit, say so honestly — not every artifact deserves deconstruction; sometimes a normal read is correct

## When NOT to use this skill

| Situation | Use instead |
|---|---|
| User said `/deconstruct-toolkit:artifact-deconstruct` (or similar) explicitly | Invoke that skill directly |
| Target is code | `sourceatlas` |
| User is debugging their own thinking | `philosophers-toolkit` |
| User wants to write new copy / docs / design | `copywriting-toolkit` / `docs-team` / `design-team` |
| Artifact < 200 words and not a structured argument | Tell the user there's not enough design to deconstruct |

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [kouko](https://github.com/kouko)
- **Source:** [kouko/monkey-skills](https://github.com/kouko/monkey-skills)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-kouko-monkey-skills-using-deconstruct-toolkit
- Seller: https://agentstack.voostack.com/s/kouko
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.