X

@cursor_ai

Steer SDK agents mid-run: Cursor SDK now has run.steer(), which adds your message to the agent's next turn; a subagent that is mid-task moves to the background and keeps working. post (https://x.com/cursor_ai/status/2107141004482793827) · 1,196❤ · 80 RT

@poteto

pstack /poteto-help: pstack v0.15.13 adds a /poteto-help skill trained on all her guides, to ask which pstack skill or poteto-mode fits a task. post (https://x.com/poteto/status/2107158163145576902) · 2,328❤ · 118 RT

Grok Bot 101 workshops: invites sign-ups for live workshops tomorrow showing how to use the bot for real work. post (https://x.com/poteto/status/2107312605429940583) · 1,008❤ · 70 RT

Template rewards: reminds that high-quality Grok Bot templates can earn rewards and asks people to share theirs in the thread. post (https://x.com/poteto/status/2107236325271404601) · 940❤ · 46 RT

@hwchase17

Agent memory cleanup: on Cognition's Devin "Dreaming", says memory needs an offline cleanup loop, not just better retrieval; open question is how inferred memories get validated; welcomes the open Agent Memory Repo standard. post (https://x.com/hwchase17/status/2107206402078851519) · 125❤ · 6 RT

Harness take: quotes David Cramer arguing generic coding harnesses like Codex will be superseded by specialized harnesses and local models; calls it a good take. post (https://x.com/hwchase17/status/2107326639231344822) · 47❤ · 0 RT

@rauchg

gdp-ts: introduces Ghosts of Departed Proofs for TypeScript, a library, linter, and AI skill where sensitive functions require compile-time proofs that the caller ran an authorization check; argues agents thrive under hard constraints humans found too costly. post (https://x.com/rauchg/status/2107119811444748555) · 2,254❤ · 113 RT

@simonw

Long-number addition test: re-ran Colin Fraser's GPT-4o experiment on adding long numbers with Qwen 3.8 27B running locally, in reasoning and non-reasoning modes. post (https://x.com/simonw/status/2107158711798218848) · 412❤ · 25 RT

@HamelHusain

Eval cadence FAQ: how often to run evals depends on cost and how saturated the eval is versus the business value of catching the error. post (https://x.com/HamelHusain/status/2107169008046772252) · 53❤ · 9 RT

Silent

@martinfowler, @karpathy, @swyx: no original posts in window

Gaps

@xAI: API not authorized to read this account's posts (skipped)

GitHub trending

tester-army/e2e (https://github.com/tester-army/e2e): next-generation e2e testing framework for web and mobile apps (TypeScript, +1,398★ today)
thedotmack/claude-mem (https://github.com/thedotmack/claude-mem): persistent context across agent sessions; captures what the agent does, compresses it with AI, and injects relevant context into future sessions (TypeScript, +534★ today)
earthtojake/text-to-cad (https://github.com/earthtojake/text-to-cad): gives your agent CAD capabilities (Python, +437★ today)
pingdotgg/t3code (https://github.com/pingdotgg/t3code): agent harness control surface to run Claude Code, Codex, Cursor, and other coding agents from mobile, web, and desktop apps (TypeScript, +485★ today)
boykopovar/AnyPS5 (https://github.com/boykopovar/AnyPS5): tool for automatically porting PS5 executables to Linux and Windows (C++, +997★ today)