---
format: "aidr-story-markdown/v1"
id: "ea3dec2c74e2931ca182450f597fec98e23df2dafcc9ee53d33a1bc2d4e0db01"
canonical_url: "https://aidr.today/ea3dec2c?lang=en"
title: "World In-Context Learning for Agentic Robot Manipulation"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-11-11T00:00:00.000Z"
category: "Frameworks"
topics: ["agent","framework"]
source_urls: ["https://huggingface.co/papers/2611.12885"]
summary: "World In-Context Learning (WICL) enables a frozen model to learn manipulation tasks through simulation rehearsal and deploy frozen Task Memory on new scenes with data isolation, as implemented in the RoboScript infrastructure."
---

# World In\-Context Learning for Agentic Robot Manipulation

> [Open the canonical story](<https://aidr.today/ea3dec2c?lang=en>)

**Published:** 2026-11-11T00:00:00.000Z
**Category:** Frameworks
**Topics:** agent, framework

## Summary

World In\-Context Learning \(WICL\) enables a frozen model to learn manipulation tasks through simulation rehearsal and deploy frozen Task Memory on new scenes with data isolation, as implemented in the RoboScript infrastructure\.

## Sources

- [Story source](<https://huggingface.co/papers/2611.12885>)