# Bounded Goal Loops, Better Review Bots, and New Computer-Use Primitives *By Coding Agents Alpha Tracker β€’ May 10, 2026* Today’s practical edge is operational discipline: Codex and Hermes goal loops only become useful when you pin them to explicit validation and stop rules. Also worth your attention: Crabbox for disposable debug loops, Peekaboo 3.0 for macOS computer use, Copilot review’s jump in usefulness, and strong Codex iOS build feedback. ## πŸ”₯ TOP SIGNAL - **Bounded goal loops are the real unlock.** After three days with Codex/Hermes goal-based agents, Jason Zhou says most people use them wrong: the loop only works when you define the objective, constraints, validation method, and explicit stop conditions up front β€” not when you say "keep fixing stuff." His deeper Codex walkthrough shows why: `go` replaces dumb programmatic looping with an LLM judge, which works well for hours-long migrations, refactors, and optimization tasks, but breaks down on multi-week work without fast, verifiable feedback. [^1][^2] ## ⚑ TRY THIS - **Turn one-off prompts into a bounded `go` run (Jason Zhou).** 1. Enable the feature: `codex features list` β†’ `codex features enable go`. [^2] 2. Give it a verifiable brief, e.g. `go "migrate codebase from JS to TS, verify screens stay exactly the same visually with Playwright"`. [^2] 3. Check status with `go`; interrupt with `go pause` or `go clear`; branch a side investigation with `side`. [^2] - **Do an alignment interview before the agent writes code (Jason Zhou; Vincent from OpenClaw).** First dump the context: what the project is, what the user cares about, what "bad" looks like, what you already tried, and the bugs it keeps missing; then let the agent ask questions before it starts. Also quantify done β€” e.g. "find 20 discrete new issues, propose fixes, push fixes to a branch, log results" β€” because fuzzy goals like "keep fixing" make the model stop early or wander. [^2] - **Move the contract into files with Go Buddy.** Run `mpx go buddy`, then `goprep` to generate a `go.md` with the request, constraints, stop rules, and loop details plus a `state.yaml` task file. Then run `go @go.md` so every loop re-reads the same contract instead of relying on chat memory; Jason shows this taking a vague game idea to a functional game with generated assets. [^2] - **Debug in disposable sandboxes, not your local machine (Peter Steinberger).** His loop is simple: ask Codex to recreate the exact failing state in an ephemeral Crabbox, verify the bug, fix it, then verify the fix. The upside: no polluted local environment and enough isolation to run 10+ sessions in parallel without slowdown. [Crabbox](https://crabbox.sh) [^3][^4] ## πŸ“‘ WHAT SHIPPED - **Peekaboo 3.0** β€” Peter Steinberger says this is the biggest release since 2.0: action-first macOS computer use, unified screenshot + UI detection, cleaner JSON across CLI + MCP, and better snapshots. His framing is the interesting part: he started it last year, but says the models weren’t good enough then; now they are. [peekaboo.sh](https://peekaboo.sh) [^5] - **Crabbox Windows terminal handling** β€” strong enough that Steinberger says Codex could E2E-fix gifgrep’s animated GIF terminal rendering. Better terminal substrates clearly matter for what agents can validate. [crabbox.sh](https://crabbox.sh) Β· [gifgrep.com](https://gifgrep.com) [^4] - **Codex/Hermes goal-based agents** β€” Jason Zhou says both now ship a `/goal` feature, and his field report is clear: most users are holding it wrong. Adoption signal: Jason ran a nine-hour migration with Codex, and Vincent from OpenClaw ran it for three days across 30 rounds. [^1][^2] - **GitHub Copilot review feels materially better** β€” DHH says the review feature, *not* the local CLI, went from roughly a 1/10 to 7/10 hit rate on real issue finding. Caveat: it still re-raises concerns that were already dismissed with a πŸ‘Ž. [^6][^7][^6] - **Codex is getting strong iOS build reviews** β€” Romain Huet says it can design screens, write Swift with GPT-5.5, run the app in Simulator without opening Xcode, and click around with computer use to test it. Omar Shahine says a single-shot app built with goals got him about 95% of the way there and felt much better than Claude Code. [^8][^9][^8] - **OpenClaw loop speedups** β€” Steinberger says caching work is making Telegram loops in OpenClaw 5-100x faster. [^10] ## 🎬 GO DEEPER - **5:36-6:55 β€” How to write a `go` prompt that actually terminates.** Short, high-signal clip on the prompt anatomy: objective, constraints, validation, and stop conditions, with examples for migrations, prototypes, and eval-driven optimization. [^2] [![I used /goals command wrong... Here are all tips & mistakes](https://img.youtube.com/vi/rIs802-bXDY/hqdefault.jpg)](https://youtube.com/watch?v=rIs802-bXDY&t=336) *I used /goals command wrong... Here are all tips & mistakes (5:36)* - **9:59-12:08 β€” Where `go` stops working, and how `MISSION.md` can take over.** Jason draws the boundary cleanly: `go` is for hours-long coding loops; multi-week goals need scheduled reruns, stored summaries, explicit metrics, and human-in-the-loop escalation. [^2] [![I used /goals command wrong... Here are all tips & mistakes](https://img.youtube.com/vi/rIs802-bXDY/hqdefault.jpg)](https://youtube.com/watch?v=rIs802-bXDY&t=599) *I used /goals command wrong... Here are all tips & mistakes (9:59)* - **Study [Crabbox](https://crabbox.sh).** The durable pattern is exact-state, disposable execution environments for reproduce β†’ fix β†’ verify loops, plus enough isolation to fan out lots of sessions in parallel. [^3][^4] - **Study [peekaboo.sh](https://peekaboo.sh).** Peekaboo 3.0 is a practical reference for action-first macOS computer use and a cleaner CLI/MCP data model. [^5] *Editorial take: the best coding-agent setups are getting less magical and more operational β€” explicit contracts, state files, disposable environments, and hard verification beat "just let the agent cook."* [^2][^3] --- ### Sources [^1]: [𝕏 post by @jasonzhou1993](https://x.com/jasonzhou1993/status/2053109541932589116) [^2]: [I used /goals command wrong... Here are all tips & mistakes](https://www.youtube.com/watch?v=rIs802-bXDY) [^3]: [𝕏 post by @steipete](https://x.com/steipete/status/2053032450138276274) [^4]: [𝕏 post by @steipete](https://x.com/steipete/status/2053329609064685740) [^5]: [𝕏 post by @steipete](https://x.com/steipete/status/2053114837698249190) [^6]: [𝕏 post by @dhh](https://x.com/dhh/status/2053088652322869472) [^7]: [𝕏 post by @dhh](https://x.com/dhh/status/2053092703152337087) [^8]: [𝕏 post by @romainhuet](https://x.com/romainhuet/status/2053008990230151621) [^9]: [𝕏 post by @OmarShahine](https://x.com/OmarShahine/status/2052933682273382618) [^10]: [𝕏 post by @steipete](https://x.com/steipete/status/2053082660562497616)