Skip to content

agent0ai_agent-zero.md

Latest commit

 

History

History
42 lines (31 loc) · 2.05 KB

agent0ai_agent-zero.md

File metadata and controls

42 lines (31 loc) · 2.05 KB
Raw
repo agent0ai/agent-zero
url https://github.com/agent0ai/agent-zero
content_timestamp 2026-05-26
time_slice 2026-05
timestamp_source web_observed_github_search_result_2026_05_26
collected_at 2026-05-26 04:41:30 +0800
source github

GitHub - agent0ai/agent-zero: Agent Zero AI framework with Linux environment, memory, tools, browser automation, plugins, projects, skills, and multi-agent cooperation.

Source: https://github.com/agent0ai/agent-zero

Raw public page capture notes

This raw-style public GitHub capture was recorded by the hourly public metadata update. Shell GitHub API freshness is not claimed for this capture; the source was a web-observed GitHub result.

Visible repository metadata

  • Repository: agent0ai/agent-zero
  • URL: https://github.com/agent0ai/agent-zero
  • Stars: 17.6k
  • Forks: 3.6k
  • Watchers: 289
  • Commits: 2,089
  • Latest release visible: v1.13 on 2026-05-05
  • Primary language / stack signal: Python agent runtime with web UI, Docker, Playwright browser tooling, skills, plugins, memory, projects, and CLI connector
  • Collection timestamp: 2026-05-26T04:41:30+08:00

Visible README / page excerpts

  • The public README frames Agent Zero as a transparent, extensible framework where an agent can use a Linux environment, terminal, code execution, files, memory, browser automation, plugins, and tools it learns to create.
  • The page exposes project-scoped skills, memory, secrets, knowledge, repositories, model presets, and agent profiles.
  • The multi-agent section says the top-level agent can create subordinate agents and receive their reports.
  • The release and README surface makes Agent Zero relevant to the Self Evolve corpus as a production-facing runtime where memory, skills, tool creation, and inspectable work history can become persistent improvement assets.

Raw capture boundary

No benchmark was run, no source clone was modified, and no private or authenticated metadata was used. This file preserves public page evidence for downstream classification, model-card analysis, public reports, and the site index.