Claude Managed Agents (platform.claude.com/docs/en/managed-agents) — an asynchronous agent harness that runs on Anthropic-managed infrastructure. Official reference covering agent definitions, environments (cloud / self-hosted sandbox), session lifecycle, SSE event streaming, deployments (cron schedules), dreams (memory consolidation), vaults (credentials), memory stores, webhooks, and multiagent orchestration.
For direct Messages API calls, see anthropic-api-core; for the self-hosted Agent SDK, see anthropic-agent-sdk; for the core tool use / Agent Skills / MCP APIs, see anthropic-api-tools-mcp、Admin API は anthropic-admin-platform
(This skill covers the operational side of Claude Managed Agents.)
Directory Structure
skills/anthropic-managed-agents/
SKILL.md
references/
getting-started/
README.md
agent-setup.md
define-outcomes.md
migration.md
onboarding.md
overview.md
quickstart.md
sessions/
README.md
files.md
github.md
memory.md
mcp-connector.md
permission-policies.md
reference.md
session-operations.md
sessions.md
skills.md
tools.md
environments/
README.md
cloud-sandboxes-reference.md
environments.md
self-hosted-sandboxes-security.md
self-hosted-sandboxes.md
vaults.md
orchestration/
README.md
dreams.md
events-and-streaming.md
multiagent-orchestration.md
scheduled-deployments.md
webhooks.md
endpoints/
README.md
agents.md
deployments.md
dreams.md
environments.md
memory-stores.md
sessions.md
user-profiles.md
vaults.md
webhooks.md
samples/
README.md
create-agent-session.md
event-streaming.md
custom-tool-and-permission-events.md
session-files.md
scheduled-deployment.md
multiagent.md
self-hosted-worker.md
vault-credentials.md
scripts/
README.md
curl-lifecycle.md
curl-resources.md
Exploration Procedure
Derive the category from the task, then identify the target page in the category's README.md:
- Find the category corresponding to the task in the mapping table below
- of that category
references/{category}/README.md（samples/scripts/(the README.md directly under it) to identify the target page - Read the
.mdof the relevant page to check the details endpointsis not a per-endpoint page buta resource-level aggregate page(one file per resource covering all endpoints in tables, with beta header values also listed in each page's Notes), so look up pages directly by resource name (Agents / Sessions / Environments / Deployments / Dreams / Vaults / Memory Stores / User Profiles / Webhooks)
Task → Category mapping
| Task | Category | Reference README |
|---|---|---|
| Want to know about agent definitions (model / system prompt / tools / skills / MCP servers), outcome design, migration from the Messages API/Agent SDK, and prototyping in the Console | getting-started | references/getting-started/README.md |
| Want to know about session creation and operations (retrieve, list, update, archive, delete), files, GitHub integration, memory, MCP connector, permission policies, and tools | sessions | references/sessions/README.md |
| Want to know about environment configuration (cloud / self-hosted sandbox), cloud sandbox preinstalled contents, security model, and vaults authentication | environments | references/environments/README.md |
| Want to know about SSE event streaming, deployment cron schedules, memory consolidation via dreams, multiagent orchestration, and webhook subscriptions | orchestration | references/orchestration/README.md |
| Want to check the endpoint specifications (request/response tables) for each resource: Agents / Sessions / Environments / Deployments / Dreams / Vaults / Memory Stores / User Profiles / Webhooks | endpoints | references/endpoints/README.md |
| Want to know typical usage patterns (agent creation through session startup, event streaming, custom tool/permission, session files, scheduled deployment, multiagent, self-hosted worker, vault credentials) | samples | samples/README.md |
| Want to know representative curl commands for lifecycle operations and for environments/vaults/memory stores/deployments | scripts | scripts/README.md |