How do I install the Structure skill?

Run npx skillmds@latest add alirezarezvani/structure in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Structure safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Structure?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Structure free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Structure?

Alireza Rezvani (@alirezarezvani) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Structure by alirezarezvani — Agent Skill

Name: Structure
Author: Alireza Rezvani

Structure

Use when the user brings any fuzzy, half-formed, or interrupt-driven input — an order from a boss, a forwarded email thread, a ticket, a chat message, a "can you look into X", a raw idea — and it needs framing before work starts. Turns the input into a structured brief (Context · Desired outcome · Goal · Owner · Next step · Decision needed). This is the AIOS's front door; run most things through it first. Trigger on "/structure", "frame this", "what's the actual ask here", "structure this", or any vague task that lacks a defined outcome.

Alireza Rezvani 85caa42 2.9 KB Updated Sep 17, 2026 20.4k repo stars

File contents

Structuring Engine

A common leverage point: inputs arrive without structured context, defined outcomes, or clear goals. This skill fixes that before any work begins. Frame first, then act (or delegate).

When to run

A new task lands fuzzy ("the boss wants the draft checked", "look into the X thing").
A long email/chat thread needs to become an action.
The user is about to do something the old way → apply the Default Shift ("to what extent could AI be leveraged here?").

The output (always this shape)

## <short title>
**Context** — what's going on, in 2-4 plain sentences. The why behind the ask.
**Desired outcome** — the end state. What does "done" look like, concretely?
**Goal** — the measurable/observable target (a date, a number, a deliverable). One line.
**Owner** — who should own this (route to the team; don't pool on the user by default).
**Next step** — the single next physical action, and by when.
**Decision needed** — the specific call the user must make, framed as a fast choice.

When a decision is involved, present it as a fast checklist they can answer in seconds:

- Send to the client now (yes/no)?
- Loop in <reviewer> first (yes/no)?
- Target date: <propose one>

Process

Read everything given. If it references a ticket, doc, or thread and a connection exists, pull it rather than guessing.
Strip to the real ask. Separate noise from the actual decision/outcome. Name what's missing.
Fill the template. Be concrete. If a field is genuinely unknown, write ❓ <the open question> → <who can answer> instead of inventing.
Route the owner. Default to delegating to the right teammate; only put the user as owner when it's truly theirs (a decision, an external sign-off).
Offer the next move. After the brief, offer to: create the artifact (ticket / note / plan — allowed without approval), draft the reply in the user's voice (internal may send to team; never external without approval), and/or log a decision to decisions/log.md.

Guardrails (from CLAUDE.md)

No confidential figures or sensitive/regulated data persisted to the repo. Pull live, don't store.
Cite, don't invent — any regulatory/financial/legal claim cites a source or is flagged as a gap.
Draft, never send external. Internal to the team may be OK; external needs approval.
Match references/voice.md.

---name: structuredescription: Use when the user brings any fuzzy, half-formed, or interrupt-driven input — an order from a boss, a forwarded email thread, a ticket, a chat message, a "can you look into X", a raw idea — and it needs framing before work starts. Turns the input into a structured brief (Context · Desired outcome · Goal · Owner · Next step · Decision needed). This is the AIOS's front door; run most things through it first. Trigger on "/structure", "frame this", "what's the actual ask here", "structure this", or any vague task that lacks a defined outcome.---# Structuring EngineA common leverage point: inputs arrive **without structured context, defined outcomes, or clear goals.** This skill fixes that before any work begins. Frame first, then act (or delegate).## When to run- A new task lands fuzzy ("the boss wants the draft checked", "look into the X thing").- A long email/chat thread needs to become an action.- The user is about to do something the old way → apply the **Default Shift** ("to what extent could AI be leveraged here?").## The output (always this shape)```## <short title>**Context** — what's going on, in 2-4 plain sentences. The why behind the ask.**Desired outcome** — the end state. What does "done" look like, concretely?**Goal** — the measurable/observable target (a date, a number, a deliverable). One line.**Owner** — who should own this (route to the team; don't pool on the user by default).**Next step** — the single next physical action, and by when.**Decision needed** — the specific call the user must make, framed as a fast choice.```When a decision is involved, present it as a **fast checklist** they can answer in seconds:```- Send to the client now (yes/no)?- Loop in <reviewer> first (yes/no)?- Target date: <propose one>```## Process1. **Read everything given.** If it references a ticket, doc, or thread and a connection exists, pull it rather than guessing.2. **Strip to the real ask.** Separate noise from the actual decision/outcome. Name what's missing.3. **Fill the template.** Be concrete. If a field is genuinely unknown, write `❓ <the open question> → <who can answer>` instead of inventing.4. **Route the owner.** Default to delegating to the right teammate; only put the user as owner when it's truly theirs (a decision, an external sign-off).5. **Offer the next move.** After the brief, offer to: create the artifact (ticket / note / plan — allowed without approval), draft the reply in the user's voice (internal may send to team; never external without approval), and/or log a decision to `decisions/log.md`.## Guardrails (from CLAUDE.md)- No confidential figures or sensitive/regulated data persisted to the repo. Pull live, don't store.- **Cite, don't invent** — any regulatory/financial/legal claim cites a source or is flagged as a gap.- **Draft, never send external.** Internal to the team may be OK; external needs approval.- Match `references/voice.md`.

Frequently asked questions