01
A defined role and runtime
Choose a runtime, a compatible model and provider, standing instructions, and the equipment the work needs.
Agent conceptsAgents
What are agents in Tale?
Project agents work on tasks alongside your team. Give each one instructions, a supported runtime, a model and tools, then follow its sandbox work and review its reports and files.
How agents show up in Tale
Configure agents for a project, give them the tools and context they need, and follow the work through its reports and deliverables.
01 Projects
Scope agents by team
Each project has its own agent configuration. Project access determines who can see the agents and their tasks.
02 Knowledge
Ground every agent in what you know
Attach the task brief and reference files, and grant tools for accessible company knowledge. Give the agent clear instructions on which sources to use.
03 Automations
Run agents in multi-step work
An automation can run a separately configured agent step in a sandbox, alongside model calls, conditions and connected actions.
04 Sandbox
Watch the agent in Files and Live
Agents that run on an agent runtime work in a sandbox — browse the workspace tree, open the file they just wrote, and watch the live browser as they click through the result.
What can agents do?
Configure reusable workers for project tasks, from research and document preparation to code changes and review.
02
Supported agent runtimes
Bind Claude Code, Codex, Hermes, and OpenClaw as the agent runtime a project agent or automation agent node runs on.
Agent runtimes03
Delegation between agents
An equipped manager agent can start other agents on ready tasks in the same project, within delegation, access and capacity limits.
Task automation04
Skills as reusable bundles
Give agents reusable instructions and supporting files as skills. Equip them for tasks such as preparing documents or reviewing changes.
Project agents05
Tools for project context
Grant the platform tools an agent needs to find tasks, read files and retrieve knowledge. Access rules still apply to the sources it uses.
Knowledge
Agents FAQ
A project agent is a reusable worker in one project. Its runtime, model, instructions and tools determine how it approaches tasks. Reports and produced files stay with the task for review.
Yes. Connect your API credentials, or use a supported Anthropic subscription with Claude Code or ChatGPT subscription with Codex for task and automation agents. An administrator configures the credentials or token broker. Chat needs API credentials; subscription availability and limits depend on your provider and plan.
Use a project agent for an outcome that needs tools, files or several turns of work. Use an automation when you can define repeatable steps, conditions or triggers.
Yes. Choose a model and provider for each project agent, subject to runtime compatibility and the models available in your organization. Configuration changes apply to later runs.
Work you can try with your team
Start with a clear brief, defined deliverables, and a review process.
See your team’s next workflow in Tale
Bring a task to a guided demo, or start a workspace and try it with your own agents.