log-inbox-boss-maxxing
Skill63 starsChanged 2 days agoMITView on GitHub
Bossmaxxing: giving out tasks to the team. A log play (docs/useCases/inbox/bossMaxxing.md): Give out tasks to the people the owner works with, wherever they are (a Discord group, an X DM, Slack, Instagram), and keep track of every job across the team: who has what, what they owe, what's blocked. Each task goes out as a brief that says what to do, why, what done looks like and by when. A live monitor stays on, so every reply is answered as it lands, and someone being set up is walked through one step at a time until it works.
- Reads credentials
In this repository82 other files it gives its agents.
- AGENTS.mdAGENTS.md
- log-discord-channel-replySkill
- log-discord-forum-postsSkill
- log-discord-group-chat-readSkill
- log-discord-meme-replySkill
- log-discord-server-replySkill
- log-discord-server-searchSkill
- log-facebook-groupsSkill
- log-facebook-marketplace-listingsSkill
- log-inbox-governor-resetSkill
- log-inbox-reactionsSkill
- log-inbox-rea-looksmaxxingSkill
- log-inbox-scope-ajobSkill
- log-inbox-show-meSkill
- log-inbox-sign-inSkill
- log-inbox-triageSkill
- log-inbox-voice-noteSkill
- log-instagram-dm-replySkill
- log-instagram-new-platformSkill
- log-linkedin-networkSkill
- log-linkedin-postsSkill
- log-reddit-contributor-maxxingSkill
- log-slack-dm-replySkill
- log-twitter-affiliate-marketingSkill
- log-twitter-collab-maxxingSkill
- log-twitter-dm-inboxSkill
- log-twitter-engagement-repliesSkill
- log-twitter-engager-dmsSkill
- log-twitter-followersSkill
- log-twitter-follow-upsSkill
- log-twitter-get-the-actionSkill
- log-twitter-manager-maxxingSkill
- log-twitter-network-maxingSkill
- log-twitter-niche-map-regionSkill
- log-twitter-notificationsSkill
- log-twitter-recruitingSkill
- log-twitter-reply-guy-funnelSkill
- log-twitter-topic-repliesSkill
- log-twitter-who-needs-youSkill
- log-web-competitor-trafficSkill
- accountToDryRunSkill
- adsPowerSkill
- agentHooksSkill
- callTimingSkill
- captureRuleSkill
- commitLogSkill
- consoleSkill
- conversationsSkill
- devlogMaxxingSkill
- discordServersSkill
- facebookFlowsSkill
- facebookInspectSkill
- facebookWarmupSkill
- flowSkill
- governorSkill
- hostSkill
- humanMouseSkill
- inspectSkill
- instagramSkill
- knowledgeBaseSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 42
Give the same access the others have: Before inviting someone to a shared tool, look at who is already in it and with which role (a …
- Distinctive: line 48
Report money as it's spent: When a step costs something (a lookup, an email check), say what it cost next to what it found, and stop to ask …
- Distinctive: line 49
The live chat starts where the owner says: If the person is on two platforms and the owner names one, answer there and point the other chat …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Bossmaxxing: giving out tasks to the team
This is the log play inbox/bossMaxxing. Before anything else, call the log MCP tool getGuide({ useCase: "inbox/bossMaxxing" }): it returns this page and makes it the session's play, so every log tool's answer says which step you are on and a send with no play loaded is refused (rules 157, 158). Then follow the steps in order.
Draft every message from that person's own thread and matt's real replies (voice with no text, search), never from memory, and never send two that read alike (sameAsSent).
Give out tasks to the people the owner works with, wherever they are (a Discord group, an X DM, Slack, Instagram), and keep track of every job across the team: who has what, what they owe, what's blocked. Each task goes out as a brief that says what to do, why, what done looks like and by when. A live monitor stays on, so every reply is answered as it lands, and someone being set up is walked through one step at a time until it works.
How the owner asks for it: "bossmaxxing", "give out the tasks", "send <person> the task in <group> and tag them", "tell <person> about everything I researched and give them an outline", "triage the jobs across the team", "walk <person> through setting up <tool>", "help <person> get set up like we did with <other person>"
Run it from a fresh context
Say: "go on docs/useCases/inbox/bossMaxxing.md for <people or group>". The agent reads this page and runs the steps.
The steps
| # | Step | Tool | Done when |
| 1 | Find each person and where the owner talks to them: the group they're in, or their DM, on whichever platform he named | chatsList { account, limit? }: the platform is the account's (or discordChats, twitterDms, slackChats, instagramDms) | every person has one chat, named as the owner says it ("the aeo group", "@someone on X"), never an id |
| 2 | Read each person's chat in full before giving them anything: what they were last told, what they owe, what they said back, and what they sent (files and sheets open through the read) | messagesRead { account, chat, limit: 120 } (or discordRead { channel }, twitterDms { with }, slackRead { chat }, instagramDms { with }) | for each person: the last task, its status and the time of their last message |
| 3 | Triage the jobs across the team: one row per person with the task, its status (done, waiting on them, waiting on the owner, blocked), the last update with its time, and the next move | the reply to the owner | the owner can see who has what in one table |
| 4 | Write the task as a brief: what to do, in order; why it matters; what done looks like and by when; what to do if they're stuck. When it carries research or a long outline, put it in a file in OUTBOX/ (plain words, no frontmatter) and keep the message short | the agent, OUTBOX/<PERSON>_<TASK>.md | the brief answers clear, why, done when, kind and how to ask (rule 109) |
| 5 | Turn on the live monitor before anything goes out, so each reply wakes the agent the moment it lands | watchersStatus {} shows each account's watcher live, and watchersStart {} starts any that isn't; then Monitor { timeout_ms: 1800000 } on a small bun:sqlite poll of ~/.logCli/log.db: every 3 s, print each new events row past the last seq whose type is message, mention or reply, with its platform, withWho, text and conversation; re-armed on expiry (rule 134). A harness with no Monitor reads inboxPending { reader } before each turn, or listens to one chat with messagesWatch { account, chat, seconds } | the person's account shows live, and the monitor prints a test line from a message that just landed |
| 6 | Claim the chat | inboxClaim { platform, account, chat, event } | this session holds the chat (rule 133) |
| 7 | Send the task. In a group, tag the person (<@their id> on Discord) and send it as a brief with the file attached. In a DM, send it one message per line. Bossmaxxing always goes over the voice gate: the owner said so, so don't rewrite a draft to pass it | discordSendLive { channel, text, brief: true, file, approvedBy: "<owner>", confirmLive: true, overGate: true }, twitterDmLive { userId, text, overGate: true }, slackSendLive, instagramSendLive (rule 137) | the platform answers sent |
| 8 | Walk them through it live, one step at a time: give step 1, ask them to say done, read what they send back, then give step 2. Ask for the one thing a setup needs from them (their email, a login). Do the owner's side yourself through the tool that holds it, such as an invite in a web dashboard through the Aside REPL | the monitor, twitterDms { with } or discordRead before each reply, aside repl for a dashboard | each step is confirmed by what they sent back, not assumed |
| 9 | Tell the owner what went out and where things stand | the reply to the owner | who got what, what they said, what is still pending, and anything only the owner can do |
What the runs taught
- The monitor reads
log.db, neverpending.jsonl. Since rule 134 every pushed message lands in theeventstable, andpending.jsonlis no longer written. A monitor tailing it stays silent while the person writes, and that's how a teammate got left waiting. A message X Chat couldn't open on the push lands as amessagewith no text beside anunopenedrow: read the chat withtwitterDms{ with, full: true } and it opens. - A task is a brief, not a chat message.
brief: trueondiscordSendLivegates it on clear, why, done when, kind and how to ask (rule 109). A brief with no "done when" or no deadline comes back as "what do you want me to do". - Give the same access the others have. Before inviting someone to a shared tool, look at who is already in it and with which role (a dashboard's members or pending invitations), and give the new person the same role.
- Ask for one thing at a time. "What email do you want the invite on" got an answer in a minute. A long setup message with the question at the end didn't.
- They write while it's going out. A send that is one message per line stops when they reply partway through (
>>> INTERRUPTED). Read what they said, answer it, then send the held lines if they still fit. - A burst of empty X Chat events right after you message is read receipts, not messages. Read the person fresh with
twitterDms{ with, full: true } before saying they wrote anything. - X DMs are checked against house style. A domain or an email address counts as a full stop. Write "your gmail" or the domain without its ending, or keep the address for the dashboard.
- The work runs in a harness, never in a browser by hand. When someone does research or lead work, they run it from a coding agent (T3 Code, Claude Code, Codex), and web pages are read through the Aside REPL. Lists come back as a
.tsv(one row per line, tab between columns), never a pasted table. - Report money as it's spent. When a step costs something (a lookup, an email check), say what it cost next to what it found, and stop to ask the owner before anything he didn't ask to pay for (rule 41).
- The live chat starts where the owner says. If the person is on two platforms and the owner names one, answer there and point the other chat to it.
- A play must be loaded before sending as the owner. A send with no play loaded is refused
playMissing, and one this page doesn't name is refusedplayNotCovering. This page is that play. - A booked call, a sale or a request for access waits for the owner. A teammate may report a client call they booked, say they plan to upsell, ask for access to one of the owner's tools and which email it's on, or share an invite link. Each of those is a meeting, money or access to the owner's tools, so nothing goes out: tell the owner what was said and in which group, and leave the answer to him. Someone new a teammate brings in is introduced the same way. Say who they are and leave the welcome to the owner.
- Once the owner says help them with it, help them. Get them to explain what's happening first ("is it not letting you in or what"), then fix it. A password to the owner's own accounts (a Google admin sign-in, say) is still the owner's to give, never the agent's.
- The boss fixes it himself. The owner's words: "the point of being a boss is that you are the one who is asked a question to solve something and you be a boss and do it yourself". Teammates often say things with no context and expect them to work. Don't hand them a list of steps to do. Find the cause through the tools and keys you hold (the repo, its
.env.local, the servers over a private network, the hosting dashboard), fix it, check it works, then tell them it's done and what to do next. Tell them to leave it alone while you work on it. - Over the gate, every time. The owner's words: "go over the gate when your boss maxxing". Plain questions to a teammate can score low on the gate and hold them up. Every bossmaxxing send goes with
overGate: true. The AI artifact check still runs. - Know where the team's tools run. Keep the map of the team's hosts, subdomains and which sign-in each tool uses in the private knowledge base, never in this page. Never point anyone at a retired host or an ended trial. Before changing a tool, check its repository's branch and uncommitted work, since a move may already be half done there.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

