Use cases

What log-cli is built to do, one page each, by platform. Each page gives the idea, the steps in order, the command and MCP tool for every step, what decides that a step worked, and the lessons that shaped it. They're here so anyone can see the parts and put them together for their own idea. Every page can be run from a fresh context: say "go on <the page>" and what the round is for.

PlatformUse caseWhat it does
XAll X use cases, and the parts they sharestart here for anything on X: the queue, the rules, and how to run a page from a fresh context
XRecruiting on X with a reply queuefinds people posting that they want work, replies to each in their language, sent slowly after you approve
XReplying to everyone posting about one topicone reply per person about their own post, backed by facts you can stand behind
XAnswering the people who engaged with your postsanswers commenters under your post, or on a fresh post of theirs
XSeeing who engaged with your postsevery like, repost, reply, quote and follow from the account's notifications, new and historical, kept locally, sends nothing
XDMing the people who liked or reposted your postranks the likers, you pick, one opening line each, sent as a dm queue
XReading the DM inboxwho is waiting on you and what they asked, sends nothing
XAffiliate marketing through your DMsfollows up every conversation about a tool you use, with your affiliate credit where it helps, and books the partner call
XChecking who really follows youa live follows check before anyone is called a follower
XFinding who needs something from you, or who mattersreads every conversation for what it's about and sorts people by need, importance and how close to money, sends nothing
XFollowing up in your own voicedrafts each follow-up from how you write to that person, gates it, and sends one message per line after you say yes
XNetwork maxing: grow the network, the niche mapgets to know everyone who messaged and new people from the niche map, each conversation about them and what they're doing, never a pitch
XCollab maxxing: turn open conversations into collabsturns the people already talking to the owner, and the missed openings, into collabs: scope, position the open-source collab, then the small ask, with money always the owner's call
XA new region of the niche map, then outbound to itadds a new world to the niche map, runs it through the pipeline, then opens on the warmest of it
XReply guy, then the DM, then the askreplies to each new person's post with a meme, opens a DM on that post, builds on what they say, then the ask, no two messages alike
XGetting people to do the thingexplains each conversation in a line, then a reply that answers them and asks for the one action (retweet, like, try it)
FacebookFrom accounts to live Marketplace listingsone local profile per town, each posting a listing, clicks counted into Twenty
FacebookListening to local Facebook groupsjoins local groups and passes the posts asking for the service to Twenty
DiscordReading a Discord group chat without Twentycatches up on a group over the last few hours with a local account, sends nothing
DiscordReplying with a meme in a Discord chatreads the chat, finds a Memelord meme that fits the moment, gates it and sends it inline
SlackReading and answering a Slack DMfinds the chat by who it is with, reads it, drafts in matt's voice, gates it and sends it one line at a time
InstagramReading and answering an Instagram DMreads the thread, drafts in matt's voice or picks a meme, gates it and sends it through the engine, never the browser
Every chatReacting to a messagereacts with an emoji on X, Discord, Slack or Instagram through the platform's own engine, claimed and counted as a write
Every chat/rea-looksmaxxing: moneymaxx like a creatorfinds a creator's most viral clips through treg, decodes the formula, posts it through your accounts and funnels the replies and DMs into your offer; also an installable skill
Every chatBossmaxxing: giving out tasks to the teamreads each person's chat, triages every job across the team, sends each task as a brief with a live monitor on, and walks people through setups one step at a time
Every chatSending a voice noteonly when matt asks: his words said in his cloned voice and sent as Discord's, Instagram's or Slack's own voice note, gated and claimed like a send
InstagramBringing a new platform up, then vendoring its clientmakes the platform, opens it in the account's browser to sign in, reads its client's calls with rea, then vendors the GitHub client that speaks them

The parts every use case is built from

  • An account in Twenty (socialAccounts): its home, its session, its status. Or a local account in ~/.logCli/localAccounts.json (log localAccount add), which wins over a Twenty row of the same name; LOG_ACCOUNTS=local leaves Twenty out.
  • An engine: the direct API where there is one (twikit for X, fbk for Facebook reads), the account's own browser otherwise. The reference's Engines section says which command uses which.
  • The platform's own answer decides: every step reads what the platform sent back, never the screen or a timer.
  • A dry run first: anything that acts on a platform does nothing without --live, and over MCP it is a separate …Live tool that needs confirmLive.
  • Pacing: each account has its own governor, local hours and limits.

A new use case gets its own page here, under its platform: docs/useCases/<platform>/<useCase>.md. The log MCP serves each page as getGuide({ useCase: "<platform>/<useCase>" }), names it on every tool the page names, and the pre-commit review asks SubQ whether a session left a page stale or work uncovered (MCP server). Pages describe the method, never the owner's own accounts, people, prices, links or files: examples use placeholders like <account>, <role link> and @someone.