What is Agent Mode?
Agent Mode lets you give Nyquest a goal and have it work toward an answer in multiple steps, using real tools. Instead of one chat completion that produces text from training data, the agent can:
- Search the web
- Fetch a specific URL's contents
- Look up live market prices
- Search your own memories from past conversations
- Read artifacts you've saved
- See the system prompt and configuration of the project you're working in
Each tool call produces an observation that the agent uses to decide the next step. The loop runs until the agent has enough information to answer (or hits a guardrail).
You see every step as it happens — the planner's reasoning, which tool was called with which arguments, what the tool returned, the running cost. When the agent finishes, the answer is appended to your chat history just like any other assistant reply.
When to use Agent Mode vs Chat
| You want... | Use |
|---|---|
| A creative answer, an explanation, a summary, a draft | Chat (faster, cheaper, no tool calls needed) |
| The current price of a stock or crypto | Agent (uses quote_lookup for live data) |
| News from the past week | Agent (uses web_search) |
| The contents of a specific URL summarized | Agent (uses web_fetch) |
| Something tying together multiple sources | Agent (multi-step reasoning) |
| To pull from your saved memories or artifacts | Agent (uses memory_recall or artifact_read) |
In short: Chat is for what the model already knows. Agent is for fresh data, your own context, or multi-step work.
What Agent Mode is not
Agent Mode is currently a separate mode you toggle into. It does not run in the background, does not act on your behalf without explicit input, and cannot perform external writes (no sending email, no posting to APIs, no modifying files). Those are deliberate scope choices for the current version.
If you want continuous monitoring or scheduled actions, those aren't built yet. If you have a use case that would benefit from them, file a feedback note.
Where to next
- Quick Start — your first agent run in 7 steps
- The 6 Tools — what each tool does and when the agent picks it
- Examples — goals that work well, goals that don't
- Costs and Limits — how long, how much, what caps apply