api-reference.md

API Reference

NexusAgent

from nexus.agent import NexusAgent

agent = NexusAgent(model="ollama/llama3")

Methods

Method	Parameters	Returns	Description
execute(prompt)	prompt: str	str	Execute a task with the agent
evolve()	—	dict	Scan workspace and build memory graph

Configuration

from nexus.config import load_config, save_config, NexusConfig

config = load_config()           # Load from ~/.nexus/config.yaml
config.model.default = "ollama/codellama"
save_config(config)              # Save to file

NexusConfig Fields

Field	Type	Default	Description
model.default	str	"ollama/llama3"	Default model
model.max_tokens	int	2048	Max tokens per response
model.temperature	float	0.7	Model temperature
memory.max_nodes	int	10000	Max graph nodes
memory.max_edges	int	50000	Max graph edges
skills.directory	str	".nexus/skills"	Skills storage
skills.sandbox_enabled	bool	True	Enable sandbox
plugins.directory	str	".nexus/plugins"	Plugins storage
plugins.hot_reload	bool	True	Auto-reload plugins
web.enabled	bool	False	Enable dashboard
web.port	int	8420	Dashboard port

Plugin Manager

from nexus.plugins import PluginManager

pm = PluginManager()
pm.load_all()
pm.call_hook("nexus_pre_execute", "Hello!")
plugins = pm.list_plugins()

Sandbox

from nexus.sandbox import Sandbox

sandbox = Sandbox(timeout=30, max_memory_mb=256)
result = sandbox.execute("print('Hello!')")
print(result.stdout)       # "Hello!"
print(result.success)      # True

Export

from nexus.export import export_skills_json, export_markdown_report, export_skill_pack

export_skills_json(agent, "skills.json")
export_markdown_report(agent, "report.md")
export_skill_pack(agent, "pack.zip")

Updater

from nexus.updater import check_version, update_skills, self_update

version_info = check_version()
update_skills()
self_update()