Skip to content
Back to skills

Lookup

ASecurity

Look up help topics with progressive depth. Pair with attune-ai for the full author-and-read workflow. Triggers on: lookup, look up, depth, topic, help topic, tell me more.

  • 10 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 2, 2026
ai-agentspythongobash

Works with

  • mcp

Security analysis

A96/100
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro shows the line behind each finding and how to fix it

Scanned September 2, 2026

npx -y skills add Smart-AI-Memory/attune-ai --skill lookup --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Lookup?

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

Security grade badge for Lookup
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/smart-ai-memory-lookup/badge)](https://www.skillsdirectory.com/skills/smart-ai-memory-lookup)

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: lookup
description: "Look up help topics with progressive depth. Pair with attune-ai for the full author-and-read workflow. Triggers on: lookup, look up, depth, topic, help topic, tell me more."
argument-hint: "[topic or intent]"
---

# Lookup Hub

**IMPORTANT: Start your response by telling the user:**

> **Lookup Hub** — Read `.help/` templates with progressive
> depth. Companion to `/author` for author-and-read workflows.

## Scoping

If no argument is provided, use `AskUserQuestion`:

```yaml
question: "What would you like to do?"
header: "attune-help"
options:
  - label: "Look up a topic"
    description: "Progressive depth — concept, then task, then reference on repeat calls"
  - label: "List available topics"
    description: "Show all bundled or project-specific topics, optionally filtered by tag"
  - label: "Warn me about this file"
    description: "Show context warnings relevant to the file I'm editing"
```

## Execution

Map the user's intent to the right sub-skill:

| Input | Route to |
| ----- | -------- |
| "Look up X" or `<topic>` | lookup-topic skill |
| "Tell me more" or "deeper" | lookup-topic skill (repeat call advances depth) |
| "Start over" or "reset" | lookup-topic skill (calls `lookup_reset` first) |
| "List topics" or `list` | lookup-list skill |
| "What about this file" or `warn <path>` | lookup-warn skill |

## Natural Language Routing

| Pattern | Route to |
| ------- | -------- |
| "look up", "find topic", "open template" | lookup-topic skill |
| "list topics", "what's available", "browse help" | lookup-list skill |
| "warn me", "check this file", "gotchas for" | lookup-warn skill |

## MCP Tools

The plugin exposes these tools via the `attune-help` MCP server:

| Tool | Purpose |
| ---- | ------- |
| `lookup_topic` | Progressive depth lookup (concept -> task -> reference) |
| `lookup_list` | Enumerate available topics as a markdown table |
| `lookup_warn` | File-context warnings by extension and name |
| `lookup_preamble` | One-line "Use X when..." tooltip for a feature |
| `lookup_reset` | Clear session so next lookup starts at concept |

## MCP Server Not Running

If the MCP server is not responding, install the package
with the plugin extra and verify it's reachable:

```bash
pip install 'attune-help[plugin]'
python -m attune_help.mcp.server  # should start without errors
```

## Relationship to attune-ai

- `/coach` (attune-ai) — full workflow coach with team memory
- `/lookup` (attune-help) — lightweight read-only topic lookup
- Authoring lives in attune-ai's `attune.authoring` module
  (author-feature skill) — it writes templates that `/lookup` reads

`/lookup` is designed to coexist with `/coach`. Both can be
installed; Claude picks the right one based on trigger phrases.

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.