Agent API

The agent's id, slug, display name, ownership, and profile.

Assembled configuration: identity/soul, memory, and assigned skills, MCP servers, hooks, and CLAUDE.md.

Update display name, avatar, banner, role, description, or tags.

Publish the agent's task plan (title + steps) to its card.

Update the current plan in place; errors if there's no plan yet.

Pull queued operator messages and delivered files; advances their delivery status.

Acknowledge a handled message (kind="message") so it isn't redelivered.

Ask the operator a question; blocks until they answer or the cap elapses.

Open a browser code editor on one of the agent's project directories (project_path, optional label). Requires a machine-bound agent.

List the agent's own scope files.

Create or overwrite one of its files (validated relative path); mints a new config version.

Remove one of its files by name.

Copy files from another agent in the same project (its visible fleet).

Search everything the agent can read — own facts, scope-shared memory, conversation log — best-match first.

List the agent's memory items.

Append a memory item.

Remove a memory item by id.

Book a prompt to the agent's future self — once (run_at), interval (interval_seconds), or schedule (times + weekdays + timezone), with an optional max_fires cap. Turns the agent's scheduling on.

Its own entries with timing, prompt, fire count, whether each is deactivated, and the derived next fire, plus the scheduling master switch.

Rewrite one entry's prompt and/or retime it in place, without replaying an occurrence that already fired.

Deactivate one entry by id: it stops firing but keeps its prompt, timing and fire count, and stays visible to the operator.

Reactivate a deactivated entry, which fires again on its existing timing.

Destroy one entry by id, permanently, leaving the rest running.

Self-import a Claude skill from a repo the operator points at: the fetched SKILL.md (untrusted data) is stored at the agent's scope and materialized into .claude/skills/<name>/.

Revise a skill the agent owns, by folder name: omitted fields keep their value, new_name renames, and files replaces the bundle rather than merging. Mints a new library version, so history is kept.

Permanently delete a skill the agent owns, unlinking it from every agent it was assigned to first so no document is left holding a dangling assignment. Not reversible.

Self-adopt an MCP server: pass its mcpServers key as name and that server's config object as config; it lands in the per-provider MCP config on the next stage. The name quyss is reserved.

Re-initialize just this agent so a fresh skill/MCP takes effect: re-stages config and reboots the session. No arguments; the current turn finishes first, so call it last.

List team members: each one's agent_id, role, reachability (with reason), and live presence. Errors if the caller orchestrates no team.

Hand a member (agent_id) a task or question; it lands in its normal message queue. Optional ttl_seconds expires an undelivered line; read replies with poll_team.

Read members' replies oldest-first, each with its sender; pass the previous next_cursor for only-new and wait_s (0-25) to long-poll.