Self-improving AI Employees for everyone
Python ai-agentsai-employeeai-employeesassistantautonomous-agentsco-employeescompanygrok-botinstinctloop-agentsmuseown-your-datapersonalself-hosted
Autonomous Agent (autonomous_agent)
| Field | Value |
|---|---|
| Category | specialized_agents |
| Plugin | server/nodes/agent/autonomous_agent/__init__.py -> _specialized.py::SpecializedAgentBase.execute_op (dispatch via BaseNode.execute()) |
| Theme color | dracula.purple |
| Icon | target (U+1F3AF) |
| Tests | server/tests/nodes/test_specialized_agents.py |
Purpose
AI agent pre-configured for autonomous multi-step operations using the Code Mode patterns (agentic loops, progressive discovery, error recovery, multi-tool orchestration). Claims 81-98% token savings when paired with the autonomous skill pack.
What is unique to this node
- Intended tool set: code executors + filesystem + HTTP, typically combined so the agent writes code to orchestrate other tools.
- Intended skills:
server/skills/autonomous/(code-mode-skill, agentic-loop-skill, progressive-discovery-skill, error-recovery-skill, multi-tool-orchestration-skill). - Frontend theming: purple dracula accent.
Behaviour
See Generic Specialized Agent Pattern. Routes through the
same SpecializedAgentBase.execute_op -> execute_chat_agent -- the
"autonomous" behaviour comes entirely from the attached skill content, not a
different execution engine. (The plugin declares a tool_description for when
it is delegated to by a team lead.)
Related
- Pattern doc:
_pattern.md - Skills:
server/skills/autonomous/ - Architecture: Autonomous Agent Creation