Group

Runtime

How an agent product executes work, delegates tasks, persists state, and requests approval.

• Agent Skills

  Track every important Agent Skills behavior independently.

• Agent Skills core and discovery

  Recognize, discover, and load reusable instruction packages centered on SKILL.md.

• Agent Skills manual invocation

  Let a user or agent explicitly select and activate a named skill.

• Agent Skills progressive loading

  Load skill metadata first and defer full instructions or resources until relevant.

• Agent Skills references and assets

  Resolve supporting reference files, templates, and assets from a skill package.

• Agent Skills scripts

  Resolve and execute optional scripts included with an Agent Skill under explicit runtime policy.

• Background agents

  Keep a run going after the operator leaves the session.

• Computer use

  Inspect and act on a visual interface through screenshots and input actions.

• Concurrent subagents

  Run multiple child-agent tasks concurrently with explicit queue and fan-out limits.

• Human approval

  Pause before a tool or write until a person confirms.

• Long-term memory

  Persist notes across sessions.

• Nested subagents

  Allow a child agent to delegate further work to another child agent.

• Per-subagent model selection

  Choose or constrain each child agent's model independently of the parent.

• Project-scoped Agent Skills

  Discover skills installed for a repository or workspace.

• Sandbox network

  Allow or deny outbound network from a tool sandbox.

• Scheduled runs

  Start a task on a timer without a human prompt.

• Structured subagent result handoff

  Return structured child status, findings, citations, files, diffs, and errors to the parent.

• Subagent approval and permission boundaries

  Control inherited grants, approval prompts, least privilege, escalation, and descendant authority.

• Subagent context isolation and inheritance

  Define and control the conversation, instructions, files, memory, tools, and secrets inherited by a child.

• Subagent delegation

  Delegate a bounded task to another agent process.

• Subagent lifecycle controls

  List, inspect, pause, resume, cancel, retry, and time-limit child-agent tasks.

• Subagent MCP access

  Allow a child agent to use MCP tools or declare its own MCP servers.

• Subagent write access

  Allow a child agent to create or modify project files.

• Subagents

  Track child-agent execution, isolation, permissions, supervision, and handoff independently.

• User-scoped Agent Skills

  Discover personal or global skills outside the current project.