# Anthropic — How Warp builds self-improving agents on Claude

- Company: Anthropic (anthropic.com)
- Announced: 2026-08-26
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude
- Record: https://forck.live/items/7415-how-warp-builds-self-improving-agents-on-claude
- Subject: Claude

Warp developed a self-improving agent framework using skills on the Claude Platform, consisting of a base skill for domain knowledge and an improver skill that uses human feedback to refine the base skill via file-based updates.

## Evidence

Verbatim from https://claude.com/blog/how-warp-builds-self-improving-agents-on-claude:

> The central technique is a self-improvement loop using skills, which are file based encodings of knowledge that keep instructions out of the raw prompt. Warp evolved a self-improving agent architecture consisting of two skills, with human feedback in between.

---

Record: https://forck.live/items/7415-how-warp-builds-self-improving-agents-on-claude
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md