An agent isn't a prompt. It's a role on your team.
Name it. Choose the model. Write the mandate. Scope the tools. Assign it to workflow steps. The same definition runs the same way on every task, in every project — no matter who kicks it off.
your own
Define the roles. Choose the models. Set the rules.
AI that compounds across your organization.
When a developer finds the right model, the right prompt, and the right tool configuration for code review — that knowledge shouldn't live on their laptop. In Spaces, it becomes a shared agent definition: available to every team, assignable to any workflow, with a traceable record on every run.
And it gets better over time. Tune the prompt based on real execution data. Swap the model when something faster ships. Tighten tool access as you learn what works. Every team using that agent gets the improvement automatically.
A full definition for every agent
Model, provider, credentials, system prompt, skills, and tool allow-list — all in one place. Click an agent in the list to see how each role is configured. Definitions are reusable across spaces and workflows.
Start from a template, make it yours
Skip the blank-page problem. Spaces ships with ready-made agent templates for common roles — each pre-configured with a system prompt, model pairing, and skill tags. Use them as-is or customize everything to fit your stack.
Reviews PRs for correctness, style, and security. Runs after every implementation step.
Claude Sonnet 4.6Writes code from specs and task descriptions. The primary execution agent.
Claude Haiku 4Scans code for vulnerabilities and compliance issues. Required for sensitive changes.
Claude Opus 4.6Reviews UI changes for visual regressions, accessibility, and design consistency.
Gemini 2.5 ProAssign to workflow steps
Each workflow step can specify which agent handles it. When a task reaches implementation, the Implementer picks it up. When it reaches code review, the Code Reviewer takes over. When it reaches deploy — a human signs off. You decide how much to automate.
Assign agents to steps. Tasks route automatically.
See every run in the thread
When an agent executes a workflow step, Spaces records a structured sequence of events: tool calls, file changes, cost milestones, and completion. It is not a casual chat log — it is an attributable trail tied to the agent definition and the task.
Full control over every agent
System prompts, model selection, skill tags, tool restrictions, scope control, and performance tracking. Everything you configure lives in one definition — visible, auditable, and reusable across your organization.
System Prompts
Define what the agent knows, how it behaves, and what standards it follows. Consistent behavior across every task.
Model Selection
Choose the model and provider per agent. Use Haiku for implementation, Sonnet for review, Opus for security — whatever fits the role.
Skill Tags
Tag agents with skills: coding, review, security, testing. Workflow steps match against skills to route work automatically.
Tool Allow-Lists
Control which tools each agent can access. Restrict file system access, limit API calls, enforce sandbox boundaries.
Scope Control
Org-wide agents available everywhere. Space-scoped agents restricted to specific projects. You control the blast radius.
Performance Tracking
Tasks completed, cost per task, error rate, average cycle time. See which agents deliver and which need tuning.
Any model, any provider
Each agent can use a different model and provider. Route implementation to Haiku, review to Sonnet, security scanning to Opus, UI review to Gemini — or a self-hosted endpoint. Credentials are scoped per agent — swap models without rewriting workflows.
Spaces
Managed execution with built-in cost tracking and credentials
Anthropic
Claude Opus 4.6, Sonnet 4.6, Haiku 4
OpenAI
GPT-4.1, o3, o4-mini
Gemini 2.5 Pro, 2.5 Flash
Self-hosted
Llama, Mistral, or any OpenAI-compatible endpoint