{"slug": "the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside", "title": "The one-page contract I make every autonomous AI agent sign (template inside)", "summary": "A developer running autonomous Claude Code agents has created a one-page 'autopilot charter' contract that defines goals, autonomous powers, guardrails, and reporting formats for AI agents. The template surfaced three silent failures in the first week of use, including a scheduler logging success while publishing nothing and a dead session that had been running for days. The developer offers the charter template and watchdog agents for free on GitHub under an MIT license.", "body_md": "You can hand an AI agent a hundred tools. The thing that actually decides whether it helps you or quietly wrecks something is a document most setups don't have: a **written delegation contract**.\n\nI run a one-person, multi-channel operation (publishing, monitoring, commerce) on autonomous Claude Code agents. Every line that runs unattended runs under a one-page charter. Here's the exact template and the reasoning behind each section — copy it, fill the brackets, and you have a management layer.\n\n```\n# AUTOPILOT CHARTER — [agent/line name]\n\n## Goal\n[One sentence. Measurable. With a deadline.]\n\n## Daily routine\n[What it does on schedule, in order. Times matter.]\n\n## Autonomous powers\n[What it may do WITHOUT asking — publish to owned channels,\nedit its own queue, retry failures, rotate content.]\n\n## Absolute guardrails\n- Facts only: no number without a live source — otherwise write `unverified`.\n- Never touch credentials, payments, or anything irreversible.\n- [Your domain-specific red lines.]\n\n## Human-only blockers\n[Logins, API-key issuance, spend approval, account creation.\nSurface these in EVERY report until cleared. Never work around them.]\n\n## Report format\n[Daily, 5 lines max. Every number verified or marked `unverified`.]\n```\n\n**Goal, one sentence.** If you can't state it in a sentence, the agent can't optimize for it and you can't audit it. \"Grow the blog\" fails both tests. \"Publish 1 search-optimized post daily; >100 search impressions by Aug 31\" passes.\n\n**Autonomous powers — written as a positive list.** The default failure mode isn't a rogue agent; it's a *timid* one that asks permission for everything, or a *creative* one that invents permissions. An explicit list kills both. Everything not listed is not granted.\n\n**Guardrails: the facts-only rule.** The single highest-value line in the whole file. My dashboards used to flatter me — an agent would report \"views growing\" from thin air. The rule that every number needs a live source (or an honest `unverified`\n\n) is what made reports trustworthy enough to act on.\n\n**Human-only blockers surface in every report.** Agents work around obstacles — it's what they're built for. But some obstacles (an expired login, a missing API key) *should* stop the line. Making them a standing checklist item means they get 20 seconds of my attention daily instead of being silently routed around.\n\n**Report format is the interface.** Five lines, facts only. If the report is long, nobody reads it; an unread report is the same as no report.\n\nBefore: agents behaved like eager interns — plausible output, invisible mistakes. After: three silent failures surfaced in the first week (a scheduler logging success while publishing nothing, a dedup filter swallowing every item, a session that had been dead for days). None were new. They'd been happening *under* the old \"just prompt it well\" setup — I only found out because reports were finally forced to be honest.\n\nThe charter template plus the two watchdog agents that enforce it (ops-monitor, qa-reviewer) are free here: ** ai-employee-management-layer** (MIT, plain markdown).\n\nThe full org — 12 role charters, memory conventions, scheduling recipes, the lessons file — is the [AI Employee Company Starter Kit](https://bluelove3.gumroad.com/l/wmtdpe) that funds the build log at [builderlog.net](https://builderlog.net).\n\nIf you're running agents unattended: what's in *your* guardrails section? Genuinely curious what other people's red lines look like — comments open.", "url": "https://wpnews.pro/news/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside", "canonical_source": "https://dev.to/moonshot_1341/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside-4lk6", "published_at": "2026-07-21 04:02:11+00:00", "updated_at": "2026-07-21 04:32:25.856222+00:00", "lang": "en", "topics": ["ai-agents", "developer-tools", "ai-safety"], "entities": ["Claude Code", "GitHub", "builderlog.net"], "alternates": {"html": "https://wpnews.pro/news/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside", "markdown": "https://wpnews.pro/news/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside.md", "text": "https://wpnews.pro/news/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside.txt", "jsonld": "https://wpnews.pro/news/the-one-page-contract-i-make-every-autonomous-ai-agent-sign-template-inside.jsonld"}}