Field noteOct 21, 20261 min read1 source

Building effective agent automations / claude.dev Blog

In one minute

Anthropic published a reference implementation for a scheduled Claude Managed Agents automation on October 8, 2026.

What changed: Anthropic adds a separate guide and reference implementation for scheduled Claude Managed Agents that monitor Slack and GitHub, track changes, and send updates.

The example reads selected Slack channels and GitHub pull requests, tracks changes between runs, and posts a brief to Slack. Claude Managed Agents is described as beta.

The configuration includes an agent, deployment schedule and time zone, memory stores, a vault, an environment allowlist, and a Slack app manifest. Setup requires a Slack app and a GitHub token. The ant apply command creates the configured resources in a Claude API workspace; Anthropic says the automation then runs on its infrastructure.

The guide recommends a separate bookmark for each source rather than a fixed time window, so a late or early run does not miss or repeat items. If a source fails, its bookmark remains unchanged and the brief should disclose the unavailable source. It also calls for confirmation from Slack before updating the reporting ledger and bookmarks, scoped credentials in a vault, read-only access where possible, and a per-run spending cap.

The source does not establish reliability, costs, general availability, or performance of Managed Agents in other deployments.

Evidence / sources

  1. 1claude.dev/blog/building-effective-agent-automations
All Field Notes

Next reads

Read next

Open the graph
  1. 01topicClaudeExplore the claude topic hub.
  2. 02topicAnthropicExplore the anthropic topic hub.
  3. 03archiveField NotesOpen the latest editorial analysis.
  4. 04source ledgerSource LedgerInspect the public source evidence graph.
Who read this page?Machine requests, hidden until opened

Loading the privacy-safe route aggregate...

Open the JSON contract