projects
Agents and settings
Each project is an Agent with one current configuration: its model, instructions, capabilities, and limits. Saved settings apply to the next run.
Agent · tools · run
Build AI agents and put them to work.
Agent Studio connects models, skills, and tools in one platform. Configure an Agent and use it in conversations, APIs, messaging, and automation.
Guide
Sign in with an account from an allowed domain.
From ideas to agents
Create, test, and run in one platform
One interface for your models
Connect the right model for each task
Cost and traces per run
Attributed to the caller that started the run
01
Configure a project
Choose a model and instructions. Add skills and tools when your agent needs them.
02
Save and run
Save the Agent settings, then run from the console, API, or connected messaging service.
03
Review the results
Return to chats, Workspace reviews and artifacts. Inspect traces and usage alongside the results.
What Agent Studio covers
agent
Agent loop
An agent can load skills, call MCP tools, delegate to subagents, and read URLs. Turn and context limits keep each run bounded.
mcp
MCP tools
Register a server once, then bind it to any Agent. Each binding can limit the available tools and override outbound headers. Project-specific OAuth is supported, and secrets are encrypted at rest.
skills
Skills
Reusable instructions written in Markdown. The model sees a short list of names and descriptions, then loads the full instructions when needed.
plugins
Agent Plugins
Bring reusable skills and MCP tools into your projects from plugin repositories or uploaded archives.
catalog
Capability catalog
Search skills, MCP tools, and agents in one catalog. An opted-in Agent adds relevant capabilities for the current run without changing its saved bindings.
chats
Chats
Talk to an agent in the console. Read PDF, text, and Office attachments without a document MCP server, and follow replies and tool calls. Closing the tab does not stop the run.
workspaces
Workspaces & Sandboxes
Enable Workspace tools on an agent to keep code, files and runtime sessions in a persistent Workspace. Execute in an isolated Sandbox, review Git changes and receive approval results in the requesting chat.
images
Images
Enable image tools in the Agent settings to generate or edit images. The Agent can edit an attached image or one produced earlier in the run using its configured image model.
artifacts
Documents & artifacts
Create reports, presentations, and spreadsheets. Edit supported attachments into new files while keeping the originals. Configured storage keeps originals and results available for download.
audio
Audio transcription & summaries
Import audio from uploads or connected tools, transcribe it with a selected model, and produce summaries in the background. Keep originals and results as private Artifacts; personal records are saved when requested.
surfaces
Slack, A2A & webhooks
Connect projects to messengers, APIs, webhooks and schedules. Identity, history, approval rules and Workspace availability differ by entry point.
cost
Cost & guards
Record Studio model-call cost by project, caller and day. Configured thresholds can alert or block new runs; external Workspace CLI usage is tracked separately by its provider.
traces
Traces & audit
Inspect recorded model runs, tool activity and usage. Tracing and audit have separate scopes and retention; administrative changes and issued-secret access are audited.
Get started with Agent Studio
Learn to configure projects, connect tools, use Workspaces and follow approval results. Check which capabilities each execution surface supports. The guide is readable without signing in.
GuideAn AI agent platform.
v0.115.1