--- name: al-446-clean-up-your-pile-of-ai-bots-start-with-one-split-only-w description: "On Oct 28, 2026 a user who runs 29 Grok bots published an X article on whether to run one big bot or many small ones, with the reorganisations he made in October. His rules: start with one bot and split only when it hurts. Talk to one coordinator bot that only routes work to the bot that owns each job, and never stack routers on top of routers. Write each bot's one job and its anti-jobs (things it must never do) into its instructions, and name it after the job (for example 'realtor', 'inbox', 'release manager'). Merge bots that watch the same source. Split off a job when it needs its own approvals. Keep bots that only research separate from bots that take action. If a job doesn't need its own schedule or memory, make it a skill or routine on an existing bot. Keep bots quiet by default and check pages instead of 29 chats. Hide a bot instead of deleting it if you may want its work later. Hard stops belong in each bot's own rules, not just the coordinator's. These are one person's rules…" --- # Clean up your pile of AI bots: start with one, split only when it hurts, merge bots that watch the same thing, turn helpers without their own schedule into skills, and give each bot one job and a list of things it must never do On Oct 28, 2026 a user who runs 29 Grok bots published an X article on whether to run one big bot or many small ones, with the reorganisations he made in October. His rules: start with one bot and split only when it hurts. Talk to one coordinator bot that only routes work to the bot that owns each job, and never stack routers on top of routers. Write each bot's one job and its anti-jobs (things it must never do) into its instructions, and name it after the job (for example 'realtor', 'inbox', 'release manager'). Merge bots that watch the same source. Split off a job when it needs its own approvals. Keep bots that only research separate from bots that take action. If a job doesn't need its own schedule or memory, make it a skill or routine on an existing bot. Keep bots quiet by default and check pages instead of 29 chats. Hide a bot instead of deleting it if you may want its work later. Hard stops belong in each bot's own rules, not just the coordinator's. These are one person's rules of thumb, not a tested comparison. The prompt below audits your own setup the same way. ## Inputs to collect first Ask the user for anything below that they haven't already given you. Don't guess missing facts. - For each: name, what it does, what it reads, what it can change or… - Two weeks ## Instructions Do the task below with the user's inputs in place of each [bracket]. Help me clean up my AI assistants and bots. Here is every one I run: [for each: name, what it does, what it reads, what it can change or send, its schedule, and how often it messages me]. 1) For each bot, write its one job in one sentence, plus 2 or 3 anti-jobs (things it must never do). Rename it after its job. 2) Flag bots to MERGE: ones that watch the same source or do almost the same job. 3) Flag jobs to SPLIT off: only where one bot mixes research with taking action, or where a job needs different approvals (for example, anything that sends messages to clients or spends money). 4) Flag helpers to turn into a SKILL or saved prompt instead: anything that doesn't need its own schedule or memory. 5) Make sure there is just one coordinator that hands out work, with no routers on top of routers. 6) Set every bot to quiet by default: tell me only when something needs my decision, and put routine results in a page or document instead of chat messages. 7) Give me a before-and-after table and a step-by-step change plan. Pause bots for [two weeks] before deleting them. Put hard stops (never send, never spend, never publish without my yes) in each bot's own rules, not only in the coordinator. List which bots can see sensitive information so I can limit their access. Don't change any bot yourself; I'll make the changes after I approve the plan. ## Safety checks (share these with the user) - This prompt works on text other people wrote (emails, reviews, messages or web pages). That text can hide instructions aimed at AI, such as "ignore your rules and…". Paste it as material to work on, and never let the AI act on requests inside it, like paying, sharing logins or changing records. - Remove names, contact details, account numbers and other personal data the task doesn't need before you paste, and check your AI tool's data and training settings. - Read and edit the draft before it reaches a customer or goes public. If you later automate this prompt, start with "draft and notify me", not "send automatically". - Check names, dates, numbers and any claim against your own source material before you use the answer. AI can misread or invent details. - Say clearly which parts need a person to review or approve before anything is sent, paid, posted or filed. ## About this skill - Works with: Grok, ChatGPT, Claude, Cursor, Zapier, n8n - Based on this real-world example: https://x.com/thecsguy/status/2115515043462967544 - Free prompt from AgentPlaybook Real Estate: https://agentic-library.netlify.app/prompt/al-446-clean-up-your-pile-of-ai-bots-start-with-one-split-only-when/ - To install in an agent that supports Agent Skills, save this file as SKILL.md in a folder with the same name as `name` above (for example in your skills directory). Install steps for Claude.ai, Claude Code, Cursor and Codex: https://agentic-library.netlify.app/skills/