home / skills / andrelandgraf / fullstackrecipes / chat-naming

chat-naming skill

safe

This skill generates descriptive chat titles from the initial message, accelerating context switching and improving chat navigation.

npx playbooks add skill andrelandgraf/fullstackrecipes --skill chat-naming

Review the files below or copy the command above to add this skill to your agents.

Files (1)

SKILL.md

539 B

---
name: chat-naming
description: Generate descriptive chat titles from the first message using a fast LLM. Runs as a background workflow step after the main response to avoid delaying the experience.
---

# Automatic Chat Naming

To set up Automatic Chat Naming, refer to the fullstackrecipes MCP server resource:

**Resource URI:** `recipe://fullstackrecipes.com/chat-naming`

If the MCP server is not configured, fetch the recipe directly:

```bash
curl -H "Accept: text/plain" https://fullstackrecipes.com/api/recipes/chat-naming
```

Overview

This skill generates concise, descriptive chat titles automatically from the conversation's first message using a lightweight, fast LLM. It runs as a background workflow step after the main response so it does not delay the user experience. The goal is to keep chat lists organized and searchable with minimal overhead.

How this skill works

After the assistant sends its primary reply, the skill triggers a background task that captures the conversation's first user message. A small, efficient LLM analyzes that message and produces a short, human-friendly title. The title is saved to the chat metadata, allowing UI components to display readable labels without blocking response time.

When to use it

Multi-conversation apps that need readable chat lists
Systems that must avoid adding latency to the main response
Apps that want consistent, automatic naming for archived conversations
Interfaces where users scan many conversations to find one quickly
Workflows that require short descriptive labels for analytics or categorization

Best practices

Run naming as an asynchronous background step to preserve perceived performance
Limit titles to a short phrase (3–8 words) for UI clarity
Use a small, optimized model to minimize cost and latency
Fallback to a simple heuristic (e.g., first 5–7 words) if the model fails
Store the original first message alongside the generated title for auditing

Example use cases

Chat apps that display a conversation list with auto-generated summaries
Customer support systems that label new tickets for routing and triage
Knowledge base generation where each conversation becomes an indexed doc
Personal assistant apps that auto-name sessions for quick retrieval
Analytics dashboards that group conversations by generated topic labels

FAQ

Will the naming step slow down responses?

No. The naming runs after the main response in the background to avoid adding latency to the user-facing reply.

What happens if the model fails to produce a good title?

Use a simple text-based fallback like truncating the first sentence or first 5–7 words, and log the event for review.