How skills and agents work in AWS Quick: ReAct loops, SKILL.md, progressive disclosure
AWS ML Blog · rss · 2026-11-10
An AWS ML Blog deep dive explains how AWS Quick — an AI service for automating tasks, analyzing data, building apps, and research — wires together agents, skills, and data, built on Anthropic's open Agent Skills specification.
- Agents: configurable AI with instructions, knowledge sources, and connectors. The built-in Quick agent has no fixed scope; custom agents (e.g., an HR onboarding assistant) can be published and shared.
- Data: spaces (curated doc collections), knowledge bases synced from Google Drive/SharePoint/S3, datasets and dashboards from Quick Sight, and connectors to Outlook/Slack/Jira/Salesforce plus remote MCP servers.
- Skills: folders with a SKILL.md (name + description frontmatter) that teach tasks. System skills generate Office docs, browse the web, run code, do deep research, transcribe AV, hand off to coding agents like Kiro — toggleable but not editable.
Agents run a ReAct-style loop and load skills via progressive disclosure: only names/descriptions are read until a request triggers the full instructions. Includes a release-notes-generator example and a full path to building and sharing team skills.
More from coding & agent
- Elastic TTS: custom text-to-speech engine coded by Gemini 4 Argon — alexanderchen · 2026-11-10
- 'Proof-of-Agent' mining: one user runs POA across Hermes, OpenClaw, Grok Bot and Muse — adamamcbride · 2026-11-10
- Vercel MCP opens to all clients via Client ID Metadata Documents, ending allowlist friction — steren · 2026-11-10
- User Lets Fable 5.2 Max 'Do Whatever It Wants' — It Keeps Working for Over an Hour — mallow610 · 2026-11-10
- Simon Willison vibe-codes a Beep Locator with Claude Code to triangulate a mystery beep using five devices — RealGeneKim · 2026-11-10
- Free MCP server vulnerable.now merges CVE data from 8 sources, no API key — Mistilziela · 2026-11-10