foreman/agent-watcher
1
0
Fork 0
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions
agent-watcher/docs/channels/routines-summary.md
bob-boat 3eda72df28 Add Channels reference docs snapshot for Layer 2 implementation 
Snapshot of official Anthropic Channels docs (channels.md, channels-reference.md,
routines summary) fetched 2026-05-06 from code.claude.com. Bob needs these to
implement the MCP Watcher in Layer 2; the channels-reference.md is the primary
implementation reference.

docs/channels/README.md cross-references the snapshot back to spec §4 and notes
the key facts confirmed: capability key, notification method, lifecycle, the
research-preview --dangerously-load-development-channels caveat, and version
minimums (v2.1.80 for channels, v2.1.81 for permission relay which we are not
using in v1).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-06 16:04:32 -04:00

1.4 KiB
Raw Blame History

Source: https://code.claude.com/docs/en/routines.md

Fetched: 2026-05-06

Routines — Executive Summary

Routines are saved Claude Code configurations that run automatically on Anthropic-managed cloud infrastructure. Each routine packages a prompt, one or more repositories, and a set of MCP connectors, then triggers on a schedule, API call, or GitHub event.

Routines support three trigger types: Scheduled (recurring or one-off timing), API (HTTP POST with bearer token), and GitHub events (PR opened/closed/updated, releases). A single routine can combine multiple triggers.

Key characteristics:

  • Run autonomously in cloud sessions with full tool access (Bash, Write, Edit, etc.) matching their configured permissions
  • Each session is isolated; GitHub-triggered routines spawn a new session per matching event
  • Connectors (Slack, Linear, Google Drive, etc.) and repositories are selected at creation time
  • Managed from the web UI at claude.ai/code/routines, the Desktop app, or the CLI (/schedule)
  • Charged against subscription usage like interactive sessions; scheduled runs count toward daily caps, one-off runs do not
  • Available on Pro, Max, Team, and Enterprise plans with Claude Code on the web enabled

Common use cases: PR review, alert triage, deploy verification, backlog maintenance, docs drift detection, library porting between SDKs.

For detailed setup instructions, see the full documentation.