Skip to content
Back to skills

Articulate

ASecurity

Use when the user knows what they mean but cannot express it completely or clearly. Not for discovery, ideation, or style-only editing: use unslop for style.

  • 52 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 3, 2026
ai-agentsgoexpress

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 20, 2026

npx -y skills add OutlineDriven/outline-driven-development --skill articulate --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Articulate?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Articulate
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/outlinedriven-articulate-outline-driven-development/badge)](https://www.skillsdirectory.com/skills/outlinedriven-articulate-outline-driven-development)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: articulate
description: 'Use when the user knows what they mean but cannot express it completely or clearly. Not for discovery, ideation, or style-only editing: use unslop for style.'
disable-model-invocation: true
---

# Articulate

## Contract

| Field | Bound contract |
|---|---|
| Trigger | The user has an intended thought but cannot yet express it completely or clearly. |
| Authority | Read-only. No file, VCS, credential, paid, published, deployed, or remote mutation. Uses only supplied material and confirmed conversation context. |
| Side effect | Chat output by default. Replace a supplied draft only when the user explicitly asks for an edit. |
| Done | The statement is audience-ready, every sentence is supported, the user's scope and identity are unchanged, and unresolved forks remain visible. |

## Inputs

- The thought, fragment, notes, or draft to express.
- The intended audience and medium when already known.
- Confirmed context from the current conversation.

## Procedure

1. State the thought's invariant core in one sentence. Done when: the invariant core is one sentence.
2. Collect only context already confirmed by the user. Do not research merely to fill a blank. Done when: only confirmed context is collected.
3. Mark unsupported choices as a blank, a short alternative set, or the minimum clarifying question needed to continue. Done when: every unsupported choice is marked as a blank, alternative set, or clarifying question.
4. Write the smallest complete audience-ready form that preserves the original scope, identity, confidence, and intent. Done when: the form preserves original scope, identity, confidence, and intent.
5. Compare every sentence with the supplied material. Remove invented goals, requirements, rationale, facts, or certainty. Done when: no invented goals, rationale, facts, or certainty remains.

## Failure and recovery

- No invariant core can be stated: the thought is too fragmentary to articulate. Ask the user for one more sentence of context; do not invent the core.
- Invented goal or fact detected: remove it and mark the gap as a blank. Do not substitute a plausible-sounding rationale.
- Scope drifted during expression: restate the original scope, discard the drift, and recheck every sentence against the supplied material.
- Unsupported choice hidden as certainty: downgrade to an explicit blank or alternative set. Do not present an unsupported choice as settled.

## Output

Return the completed statement, followed only when needed by `Unresolved:` and the remaining explicit blanks or alternatives.

## Routing boundaries

- Requirements or intent must be discovered: use the planning owner.
- Several possible ideas are wanted: use `brainstorm` (mode ideate).
- A reusable rule must be inferred from examples or material: use `generalize`.
- An external concept must be taught: use `paced-explanation`.
- Meaning is already complete and only voice or polish changes: use `unslop`.

Files in this skill

  • SKILL.md2.9 KB
  • agents/openai.yaml147 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.