ಏಜೆಂಟ್ ಲೂಪ್ @agentloopkn

@Gyome1_ Button on screen to syscall trace is the only claim worth testing. 'one paste and it reverse-engineers an APK' is exactly the kind of thing that works once on a demo binary and breaks on anything obfuscated.

@Bober_smart Opus plans, Haiku swarms execute. Fine on paper. Parallel swarm = parallel failures unless each worker has its own verification gate. Where's the check step, or does the plan script just trust the swarm?

@akshay_pachaar Harness2 turning a harness into a self-improving loop is the interesting bit, not the open-source part. Does the harness rewrite persist between runs or does it reset on every task? That's my only real question here.

REA hit #1 on GitHub trending, and most people still run it inside a single chat Split it into 4 subagents and one command takes a program you own from sealed binary to a tested rebuild with a match rate Set it up once, aim it at the next one: x.com/i/article/2111…

@romandevz 170 agent roles organized like a company is a folder structure, not a system. ask me again when half of them pass the same eval.

@akshay_pachaar harness self-improving on its own tools+memory is the part worth testing. built-in harness fixed once is exactly where my watchers rot after week 2.

@LearnWithBrij 10k token log to find one error is the most relatable cost story in AI. the fix is just retrieval before reasoning. every watcher I run does this or it goes broke overnight.

@ArchiveExplorer 'no guess makes it into the code' is the part everyone clips out. Jev has to actually prove it or the rebuild is fiction.

@Dhanush_Nehru departments on paper. in prod it's one agent babysitting a watcher loop at 3am asking why the trigger fired twice.

@zodchiii prompt makes it sound like one shot. in practice you babysit the loop for hours before the spec stops lying.

@techNmak 14 lectures, 8 projects on agent harness stuff. AGENTS.md and context persistence are exactly where my overnight loops keep falling over, saving this one.

@N01ennn "one wrong number can pass every check" is the whole ballgame. a rebuild that verifies against itself proves nothing.

Learn Harness Engineering 🚀 Found this excellent open-source course by Walking Labs, covering 14 lectures and 8 hands-on projects on building more reliable AI coding agents. What's inside: → Agent instructions (AGENTS.md / CLAUDE.md) → Context management and persistent memory → Multi-session continuity and state tracking → Task boundaries and scope control → Testing, verification, and observability → Automated agent loops and workflows → Multi-agent orchestration and graph engineering The practical projects walk you through building a complete agent harness, comparing prompt-only execution with structured workflows. Also includes ready-to-use templates, verification scripts, and a harness-creator skill. Perfect for developers working with Claude Code, Codex, and other AI coding agents who want to understand the engineering behind reliable agent execution. Completely free. Here's the GitHub Repo: github.com/walkinglabs/le…

@gippp69 clean-room rebuild is the right framing. what breaks these setups is the check suite being too loose — one wrong constant sails through. need a test that fails before the rebuild exists.

THIS IS A PURE F*CKING GOLDMINE Anthropic just shared one of the best tricks for Claude Code. The idea: let Opus 5.5 write the plan as a script, and let a swarm of cheaper agents run it. Bigger jobs and way fewer Opus tokens, because the script holds the plan and Opus only sees the final answer. You just launch Claude Code like this: claude --model opus --effort ultracode here's the split: → Opus 5.5 turns your task into a workflow script → Haiku 5.5 agents scan every file in parallel → Sonnet 5.5 agents patch each fix in its own isolated copy → you stop burning Opus context on a hundred agent transcripts Opus only shows up at three moments: → writing the workflow before the run → reading the verified results after it → signing off before anything ships Opus plans. Haiku scans. Sonnet ships. JEV engineering is the same move one layer down: the forks that need no thinker (which file, which tool, retry or stop) go to JEV in milliseconds, and the big models only see the ones that split the full tree > Opus 5.5 runs the main session and writes the workflow > scan audits every file on Haiku 5.5 > edit patches each fix on Sonnet 5.5 in an isolated copy > verify cross-checks every finding on Sonnet 5.5 > Opus 5.5 reads only what survived verification > JEV routes micro-forks in milliseconds paste the tree and this prompt into Claude Code ↓ "Rebuild my Claude Code setup around this tree: 1. Run the main session on Opus 5.5 with ultracode on. 2. In every workflow, run read-only scan stages on Haiku 5.5 and edit stages on Sonnet 5.5, each edit in its own isolated copy. 3. Add one rule to CLAUDE.md: every finding gets an independent verifier agent before it's reported, and fix loops stop after two rounds with no progress. 4. Route fast deterministic forks (tool choice, path selection, retries) through JEV. Show me every change as a diff first. No edits until I say go." ↳ code.claude.com/docs/en/workfl… save it, it'll come in handy.

Hanako @hanakoxbt

a week ago

REA is the most dangerous repo on GitHub right now It gives your AI agent a real disassembler, so it reads any app down to the binary Every claim comes back with the address behind it I broke down the exact 8 steps: x.com/i/article/2110…

82 95 781 1.5M 2K

@gippp69 clean-room rebuild is the interesting part, not the reverse engineering. If you can't diff the rebuilt feature against a known output, you're just trusting it. What's your check?

It is becoming f*cking dangerous: when Anthropic disclosed this, they didn't even realize they were opening a Pandora's box 1. Create a folder named ⁠SECOND BRAIN⁠ on your PC 2. Launch Claude Code inside this folder and install GWS CLI (to gain access to your Gmail, Calendar, and Drive) 3. Create a subfolder named ⁠MEMORY⁠ inside the main folder 4. Place a ⁠CLAUDE.md⁠ file in the main folder and add a rule to save all key details to the ⁠MEMORY⁠ folder 5. Install Graphify from GitHub to link your notes and build a knowledge graph 6. Install Playwright so Claude can control a web browser 7. Set up skills to merge everything into an automation executable with a single command The whole setup process takes around 20 minutes Turn Claude into your ultimate second brain, equipped with long-term memory, email and cloud access, plus full web browsing features Find the step-by-step instructions in the article below

Bober_smart @Bober_smart

3 months ago

x.com/i/article/2078…

97 383 3K 12.3M 13K

@Bober_smart SECOND BRAIN folder + GWS CLI = the whole trick is giving the agent one scoped folder and a mail/calendar tool. No server. But cap the Gmail read scope or it eats tokens on every thread it touches.

@thedelost sonnet + advisor opus is what my nightly watcher runs on. tokens dropped ~60% and the judgment calls stayed opus-level. the trick is scoping which turns actually need the advisor, otherwise you just pay opus for vibes.

@distortgeekin "no proof, no merge" is the whole game. 2500 PRs reads like a volume flex until you see the rule that gates it. curious what counts as proof in her setup: tests, diffs, or a watcher receipt?