Anthropic shares reference implementation for reliable scheduled agent automations
cfregly · x · 2026-11-01
Anthropic's Lance Martin and CJ Avilla published "Building effective agent automations", a walkthrough of a reference implementation built on Claude Managed Agents (beta): a scheduled agent reads sources like Slack and GitHub repos on a cron cadence, tracks changes since the last run via bookmarks and a ledger, and posts updates to Slack.
Key points:
- The project ships as config files: agent.md (model/tools/instructions), deployment.md (schedule, timezone, budget), environment.yaml (network allowlist), memory store preferences/state, and vault.
- It focuses on common failure modes: agents can silently lose access to a source or ignore user preferences.
- Practical guidance: keep the bookmark when a source read fails and name the failure; deployment-run records track session-creation failures separately from session events, and some stops leave no record, so check expected runs too.
- The poster adds advice: visible failure disclosure plus manual rerun is enough for low-stakes personal briefs; add extra checks only when people depend on recurring coverage — and drill three failures (failed read, failed start, unclear delivery recorded as maybe-posted).
More from coding & agent
- 'Am I in a cheap simulation?' Site uses Bitcoin blocks to prove reality to AI agents — voooooogel · 2026-11-01
- Dev tool UIs are converging: Vercel, Supabase, Cloudflare and ChatGPT sidebars look alike — ColleenMBrady · 2026-11-01
- 9.1B tokens in 105 days: a power user's AI usage stats — Angaisb_ · 2026-11-01
- Anthropic publishes Skill authoring best practices: keep SKILL.md under 500 lines — bibryam · 2026-11-01
- Dev builds entire Three.js city with no 3D models using Codex and GPT-6 Astra — OpenAIDevs · 2026-11-01
- Compiler-backed coding agent Benzi hits 82.6% on SWE-bench Verified for $9.26 — DonkeyTheKing · 2026-11-01