XAgent — Autonomous LLM Agent for Complex Tasks
Autonomous agent by OpenBMB with outer/inner loop planning, tool execution, and a human-in-the-loop GUI for long-horizon tasks.
XAgent
XAgent is an experimental open-source autonomous agent designed to solve long-horizon tasks via a two-loop architecture: an outer loop for high-level planning and task decomposition, and an inner loop for tool-calling execution and error recovery.
Key Features
- Dual-loop design: PlanAgent (outer) + ToolServerInterface (inner)
- Rich built-in toolset: file I/O, shell, web browser, Python REPL, search
- Human feedback mode: approve/reject individual steps via web UI
- Docker sandbox for safe code execution
- Sub-agent delegation for parallel workstreams
- Plugin system for custom tools
Quick Start
git clone https://github.com/OpenBMB/XAgent
cd XAgent
docker compose up
# Access UI at http://localhost:5173
Install via ai-supply
npx ai-supply add xagent-autonomous-task-solving
Curated mirror of the open-source XAgent (Apache-2.0). Get it from the source.
Compromise signals — malicious or tampered code (leaked secrets, backdoors, a dropped executable) — reduce the score, and known dependency CVEs carry a bounded penalty (they warrant review but never QUARANTINE — update the dependency to clear). Other dangerous-by-capability traits are risk surface, expected for some capabilities. Every finding is mapped to its OWASP control below.
Findings mapped to the OWASP Top 10 for LLM Applications (2025) and the OWASP Machine Learning Security Top 10. Expand any flagged control for the exact findings — compromise reduces the score; expected/risk-surface do not, except a known CVE, which carries a small bounded penalty (high/critical → Review).
The same gate an agent runs before installing (POST /api/v1/trust/xagent-autonomous-task-solving/check). Click a policy:
Consume XAgent — Autonomous LLM Agent for Complex Tasks programmatically. Authenticate with an API key or session — see Authorize an agent.
# Agents: CHECK BEFORE YOU INSTALL (no auth) — score, grade, level, capability manifest
curl https://ai-supply.store/api/v1/trust/xagent-autonomous-task-solving
# Gate against your org policy (returns { pass, violations })
curl -X POST https://ai-supply.store/api/v1/trust/xagent-autonomous-task-solving/check \
-H "Content-Type: application/json" \
-d '{"minGrade":"B","denyPermissions":["shell"],"denyUnknownEgress":true}'
# CLI
npx ai-supply add xagent-autonomous-task-solving
# REST (install → download)
curl -X POST https://ai-supply.store/api/v1/listings/xagent-autonomous-task-solving/install \
-H "Authorization: Bearer $AIM_KEY"
# MCP tool
install_listing({ "slug": "xagent-autonomous-task-solving" })OpenAPI spec →Curated mirror — latest upstream source. See the repository for tagged releases.