Agents

Agents mode hands a long-running task to an autonomous Copilot SDK agent attached to a topic or chat, and lets you follow its progress in a workflow-style timeline. It's opt-in and off by default.

An agent session — the task at the top, then a workflow-style timeline: reasoning, tool calls, and the assistant's answer with suggested follow-ups.

Enabling agents

Agents mode isn't part of the default install — it lives behind an agents extra that bundles the native Copilot CLI runtime:

uv sync --extra agents                 # adds github-copilot-sdk
uv run --extra agents precursor --dev  # …or run the dev stack with it

Installing the extra only makes the runtime available. Agents stay disabled until you turn them on in Settings → Agents, where the UI also reports whether the runtime resolved on your platform.

~90 MB native runtime

The github-copilot-sdk wheel bundles the native runtime binary (~90 MB download, ~145 MB on disk) — which is exactly why it's an opt-in extra rather than a default dependency. See installation.

Following a run

Give an agent a task prompt and it works autonomously, streaming its steps into a timeline you can watch. Long runs are windowed client-side so even a very long transcript doesn't mount thousands of nodes at once.

• Tool calls are visualized inline, so you can see what the agent did.
• Permissions — the agent surfaces permission prompts for actions that need your approval.
• Usage — token/usage accounting is tracked per session.

Unread badges & notifications

Agent sessions track unread activity just like topics and chats. When a background or scheduled agent produces a new reply while you aren't looking, its row in the Agents list is bold with an unread count, and — when notifications are enabled and the window is unfocused — a browser notification fires. Opening the session clears its badge.

Scheduling agents

An agent can carry its own recurrence (an AgentSchedule) so it re-runs its stored task on a cadence — the first-class equivalent of a scheduled topic that nudges an agent. Each due tick either replays the task on a fresh transcript (clear_context) or sends a follow-up in the existing conversation. A run is skipped (not failed) while the agent is mid-turn, archived, or task-less.

You can also drive an agent from a scheduled topic with slash directives:

• /agent <uuid> /clear <follow-up> — reset the transcript (same uuid), then send a follow-up.
• /agent <uuid> /run [extra] — reset, then replay the agent's own task prompt (plus an optional one-off extra), keeping instructions in one place.

See the scheduler for how these directives — and /guard gating — fit together.