Getting Started¶
ai-assist integrates with LLMs via the Anthropic Messages API, MCP (Model
Context Protocol) servers, Agent Skills, and a temporal knowledge graph to
provide intelligent monitoring, querying, and workflow automation.
Prerequisites¶
Install¶
Install uv if you don't have it:
Then install ai-assist:
Configure authentication¶
Copy the example environment file and fill in credentials for one of the methods below:
Enterprise / company Claude access via Google Cloud.
Full walkthrough: see VERTEX_AI_SETUP.md in the repository.
Personal use with a free tier from Anthropic.
Configure identity¶
Set up your identity so the assistant knows who you are and adapts to your preferences:
Edit ~/.ai-assist/identity.yaml to configure your name, role, organization,
and communication preferences:
version: '1.0'
user:
name: 'Your Name'
role: 'Your Role'
organization: 'Your Company'
context: |
Describe your work context, team, and priorities here.
assistant:
nickname: 'Nexus'
preferences:
formality: 'professional'
verbosity: 'concise'
emoji_usage: 'moderate'
The user.context field is especially important — it is injected into the system
prompt and gives the assistant awareness of your team, tools, and workflows.
Tip
Changes to identity.yaml are auto-reloaded — no restart needed.
📖 Full identity guide: Identity Configuration
Run¶
uv run ai-assist # interactive mode (default)
uv run ai-assist /monitor # start monitoring mode
uv run ai-assist /query "..." # one-off query
uv run ai-assist /run workflow.awl # execute an AWL workflow
Next steps¶
- Knowledge Quick Start — start capturing and querying knowledge.
- Scheduled Actions — set up automated monitoring and one-shot future actions.
- Personal Skills — install and build agent skills.
- Plugins — bundle skills and MCP servers via marketplaces.
- AWL Specifications — the Agent Workflow Language reference.