plusxbt✓

RT @plusxbt: I don't understand why everyone isn't using this yet: Claude Code can audit your CLAUDE.md and catch rules meant for older models 1. Update Claude Code to v2.1.283 or newer: claude update 2. Open your project and run /doctor prompt-audit 3. It reads CLAUDE.md, AGENTS.md, rules, skills and commands in .claude/ and ~/.claude/ 4. It flags instructions aimed at older models 5. It finds references to files and commands that don't exist 6. It catches files that contradict each other 7. You get a report with proposed edits, nothing changes until you ask Claude to apply them One folder only: /doctor prompt-audit .claude/skills/deploy Anthropic's docs: keep each CLAUDE.md under 200 lines, and if two rules contradict each other, Claude may pick one arbitrarily Full CLAUDE.md and folder setup in the article below

RT @plusxbt: It is becoming f*cking dangerous: one prompt turns Claude Code into a FULL crypto trading agent that switches tactics when the market flips Zero code. Copy, paste, fill in exchange, pairs, timeframe, capital: Time: 1 min to paste Action: drop into Claude Code One-strategy bots trade a flat day and a dump the same way. This one does not Regime Detector: labels each bar SIDEWAYS, CHOPPY or DUMP via ADX+ATR, swaps playbook Kill Switch: equity falls 15% below peak, positions close, halt survives restarts Backtest Gate: 3+ years with fees, slippage and unseen data, blocks any strategy that lost or fell >15% Nightly Review: writes a tweaks report overnight, nothing changes without your OK Paper Mode: sim orders on live prices, no real money until you set LIVE=true Old way: code and test every rule by hand Now: one prompt, Claude Code builds the bot, tests risk, shows diff before running Passed backtest = survived the past, not a profit guarantee. Stay on paper until you trust it Full prompt in the reply below, bookmark this

Prompt (paste into Claude Code, fill in the brackets): Build a crypto trading bot in this folder. Show the plan before writing any code Settings: Exchange: [EXCHANGE] Pairs: [BTC/USDT, ETH/USDT] Timeframe: [1h] Capital: [AMOUNT] 1. Regime detector: each bar, classify market SIDEWAYS / CHOPPY / DUMP via ADX + ATR over 50 bars. Playbook per regime: mean reversion sideways, smaller size or skip choppy, no new longs in dump 2. Risk before EVERY order: max 1% equity per trade (size = 1% equity / stop distance), stop-loss required, max 3 positions, daily loss 3% -> no trades until next UTC day 3. Kill switch: equity 15% below peak -> close all, save HALT flag + peak to disk, stay halted across restarts until I delete the flag; 15% default in config; test halt survives restart 4. Backtest gate: each strategy on 3+ years with fees, slippage + unseen out-of-sample window. Tune params in-sample only; refuse if loss or drawdown >15% in-sample OR out-of-sample; print both 5. Paper default: simulated orders on live prices, no real money. LIVE false unless .env has exactly LIVE=true set by me by hand; never set it yourself, in .env, code or defaults; test no real order fires without it. .env in .gitignore, .env.example with empties; never open or cat .env; paper loads no keys, public data only. Keys from .env live only, never print/commit, redact from logs and report.md. Trade-only keys; if API exposes permissions, refuse when withdrawals enabled 6. Nightly review: cron claude -p, --allowedTools reads trades/logs + writes report.md only; no --dangerously-skip-permissions, no shell, no .env. Writes mistakes + max 3 param proposals. Never applies; I approve; approved changes re-run tests + backtest gate 7. Tests for all risk rules + full diff before running anything

RT @plusxbt: THIS IS AN ABSOLUTE GOLDMINE Anthropic documented one of the strongest tricks in Claude Code The idea: Sonnet 5.5, Haiku 5.5 and Opus 5.5 in one setup, and each model only gets the work it's BEST at Opus stays on the hard calls, and it typically costs LESS than running Opus the whole time One command starts it: claude --model sonnet --advisor opus Who does what: > Sonnet 5.5 drives the main session and takes the routine work: writing and fixing code > Haiku 5.5 runs as read-only subagents (search, grep, docs), each in its own context window, sending back only a summary > Opus 5.5 is the advisor, called at decision points, and it sees the FULL conversation including every tool call and result > Opus isn't called every turn, so routine turns run at Sonnet prices, not Opus prices Claude decides when to call Opus, and it TENDS to do it at three moments: > Before it commits to an approach > When an error keeps coming back > Before it calls a task done Sonnet codes. Haiku scouts. Opus advises The math: per token Opus costs 2x Sonnet and 40x Haiku, so every routine job kept off Opus runs at half the price or less The full tree: Sonnet 5.5: main session, writes and edits code Search: read-only Haiku 5.5 subagent that maps the repo Grep: read-only Haiku 5.5 subagent that finds call sites Docs: read-only Haiku 5.5 subagent that reads API docs Opus 5.5: advisor that sees the whole conversation and every tool result when called Paste the tree plus this prompt into Claude Code Set up my Claude Code project to match this tree: 1. Make Sonnet 5.5 the main model and Opus 5.5 the advisor (advisorModel in settings) 2. Add three read-only subagents in .claude/agents/ (model: haiku, tools: Read, Grep, Glob): search (repo map), grep (call sites), docs (API docs) 3. Add one rule to CLAUDE.md: consult the advisor before committing to an approach, when an error repeats, and before calling a task done Before touching anything, show me the full diff and wait for my OK Bookmark this setup, you'll want it every session Step 3 lives in CLAUDE.md, the full CLAUDE.md guide is in the article below

RT @plusxbt: I don't understand why everyone isn't doing this yet Anthropic documented a Claude Code feature where the STRONGEST model only kicks in at key moments instead of running the whole time The setup: Agent Teams in Claude Code Rather than one session doing everything, the work splits across a team: Lead (Opus 5.5, high effort): the main session, creates the tasks and runs the team Teammates (Sonnet 5.5, medium effort): each owns its own files, UX or backend Advisor (Fable 5.1): consulted before picking an approach, on a repeated error, before done Shared task list: teammates claim tasks, file locks stop two of them grabbing the same one Start the lead from the terminal: claude --advisor fable Then tell it: Spawn three teammates, one on UX, one on the backend, one playing devil's advocate. Use Sonnet for each teammate Paste this to set up your workspace: In settings.json, add CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 to the env block, set effortLevel to high Define roles in .claude/agents/ with model sonnet and effort medium, plus a devil's advocate role Give each teammate its own set of files to avoid overwrites Add a TaskCompleted hook that blocks a task from being marked done if checks fail Show me every config change before applying it The CLAUDE.md map your agents read, full guide in the article below >

RT @plusxbt: Build an absolute monster. Merge six Claude Code pieces into one system, it works the NIGHT SHIFT while you sleep Opus 5.5 + subagents + routines + web search + Playwright + CLAUDE.md Six pieces. One system. Nobody at the keyboard Opus 5.5 -> the main model that plans every task Subagents -> independent workers in .claude/agents/, each with its own role and context, returning only a summary Routines -> scheduled cloud agents on a cron, the system runs at night while your laptop is closed Web search -> scans public pages and forums for real complaints Playwright -> opens a real browser, clicks through every screen, catches bugs CLAUDE.md + memory files -> the map the agent reads on launch, every step and cost logged Give it one goal before bed: find a real gap people complain about and have a working prototype by 7 AM One hard rule: nothing spent, nothing published, nobody contacted without your yes Start the routine at 11:40 PM and go to sleep. Here is how one night can look: 12:10 AM: web search pulls ~1,900 public posts with the same complaint, gig drivers tracking income across 3 apps 1:00 AM: it maps the paid tools people already use and the 2 complaints they keep repeating 2:30 AM: Opus 5.5 plans 9 screens and one database without asking a question 4:10 AM: subagents write the code and a landing page draft 5:45 AM: Playwright clicks through every screen and logs bugs 7:20 AM: the memory log holds every step Morning: a working prototype and a landing draft waiting for your yes, $0 spent beyond tokens, nothing published A subagent with no payment or email tool physically CANNOT buy or send anything. The rules in CLAUDE.md apply to every session The biggest mistake: treating AI like a chat for answers. Give yours a GOAL, a DEADLINE and HARD RULES You write zero code. Playwright does the clicking. You're asleep the whole time Everyone is waiting for AGI. The night shift already exists Bookmark this. Full breakdown of the CLAUDE.md map in the article below

RT @plusxbt: It is becoming f*cking dangerous: when Anthropic shipped CLAUDE.md, they didn't even realize they were opening a Pandora's box 1. Make an empty folder on your computer, call it SECOND BRAIN 2. Open Claude Code inside it 3. Add subfolders: MEMORY/, notes/, people/, projects/ 4. Drop a CLAUDE.md in the root with one rule: save every key fact to MEMORY, one topic per file 5. Add @-links in CLAUDE.md so key files load into context every session 6. Run claude mcp add to connect Playwright MCP, now Claude controls a real browser 7. Build a skill in .claude/skills/ so one command runs the entire routine Folder setup: about 5 minutes Your Claude becomes a second brain: it keeps every key fact between sessions, loads your main files by itself and drives a real browser Full guide in the article below

RT @plusxbt: Complete madness. Set up a digital chief of staff, give it one strict rule: "find me $60,000 a month in contracts or you are fired," go to sleep, and a ready plan is on your desk by morning Routines in Claude Code run on a schedule in the cloud while your laptop is closed You pick the goal. The agent grinds through every step. You just say yes or no The rule itself: "$60,000 a month in new contracts. public pages only. leave the price empty. nothing goes out until I say yes" It reads that from CLAUDE.md at the start of EVERY session, and a subagent with no email tool physically cannot send a thing How one night can look: -> 8:00pm the routine starts and splits the job between subagents, you write zero code -> 9:00pm 380 physiotherapy clinics in Denver pulled from public directories, chains removed -> 11:00pm the remaining sites read one by one, 30 targets kept -> 1:30am 30 personalised drafts written, sent: ZERO -> 3:00am a one-page offer ready, price box left empty -> in the morning you type $2,000 a month: 30 x $2,000 = $60,000 a month of potential, not revenue The best part is the empty price box: the number a client sees comes from you, not from a model Make it a routine for every Monday at 8pm and the whole night replays on its own 380 clinics -> 30 targets -> 1 offer -> $60,000 a month of potential -> 0 sent without your yes The clinics in your city sit in public directories. The only question is whose agent reads them first Bookmark this Full folder structure and the CLAUDE.md the agent reads every night, in the article below

It is becoming f*cking dangerous: AI agents in Claude Code can turn a stack of research PDFs into an interactive physics lab you open right in the browser Hand 10 subagents a pile of 48 papers and ONE brutal goal: a browser lab where anyone can rerun the experiments on their own Choose a flow model, move the sliders and watch the physics react live Flip algorithms mid-run, open any formula or script and read the numbers right next to the chart One job per agent: > analysts -> mine the PDFs for equations, assumptions and test setups > coders -> turn each equation into code that runs > auditors -> hold every curve up against the paper it came from > UI agents -> give each experiment its own control panel The trap: a gorgeous animation can still be dead WRONG on the math That is why auditors check that mass is conserved, the time step stays stable and the output hits the paper's reference values before anything ships Next to every run you see: which claim it tests which constants it uses how far it drifts from the paper's numbers what nobody has checked yet Some methods will match the paper, some will stay flagged for review, and that is exactly the point Put the PDFs in your project folder, open Claude Code and paste this prompt: Build a browser lab from these papers: 1. Pull every algorithm, model and test parameter, cite the page for each and list what is missing 2. Give each method its own isolated subagent, feed them all the same inputs and score them with one shared set of metrics 3. Run a separate review pass where auditors log bugs, and anything not yet checked gets a WARNING badge 4. Add controls to change parameters, compare methods and open the source code Start with ONE paper and ONE experiment, add agents only after it checks out Ask for a single HTML file so anyone can open the whole lab in a browser The 5-minute folder and CLAUDE.md setup for your agents is in the article below

RT @plusxbt: THIS IS AN ACTUAL F*CKING GOLDMINE YOU'RE PAYING OPUS TO GREP YOUR REPO. think of claude code as a team. opus 5.5 is the senior: $4 per 1m input. haiku 5.5 is the intern: $0.10. right now your senior does the intern's job. steal this: claude --model sonnet --advisor opus 01 search → haiku maps the repo 02 grep → haiku hunts call sites 03 audit → haiku in its own worktree 04 sign-off → opus at 3 moments each haiku works in its own context. hands back only a summary. 3 files in .claude/agents. model: haiku. sonnet builds. haiku swarms. opus reviews. how a $360/yr app goes to $0 ↓ bookmark this before your next claude code session.

THIS IS A PURE F*CKING GOLDMINE 1M tokens in. 100K out. all on Opus 5.5: $6.00. same run, cache + batch + routing: $0.20. The idea: stop paying Opus for Haiku work. 4 rules: 1. cache: Opus reads cached at $0.20, not $4 2. batch: 50% off every token, stacks with cache 3. route: Haiku 5.5 is 40x cheaper on input 4. advisor: Opus only at decision points the real leak: paying Opus on every turn. $6.00 → $0.20. about 30x. rule 4 is one flag. post below ↓

It is becoming f*cking dangerous: Oracle wanted $8.8B from Google. for ~11,500 lines of Java API. Supreme Court: fair use. 6 to 2. now your agent can read an app you own. no source code. with proof. The idea: REA reads the panel on disk. Opus 5.5 writes the spec. the whole engine: 5 settings. → cut the frame 30 to 70% → cut each half again → stop after 5 cuts → re-cut on every beat every cell is u v w h. REA reads. Opus writes. you build. behavior crosses over. code doesn't. save this before you renew your next plugin ↓