Why agents thrive on git, bash, and grep: someone designed them very carefully
vboykis · x · 2026-09-26
- vboykis answers her own question: agents succeed on primitives like git, bash, and grep because someone thought very hard about how those tools should be good.
- The thread argues that "languages don't matter anymore because agents do it" is false—the design quality of primitives underpins agent reliability.
More from coding & agent
- Explosion launches Ellf directory: 52 NLP templates plus multi-agent annotation workflow — _inesmontani · 2026-11-11
- Meme: 'Stochastic AI Agent Architecture' Is Just a Markov Chain in Disguise — PrasoonPratham · 2026-11-11
- OpenCode adds Gemini 4 Carbon and Claude Fable 5.2 support — dean_rie · 2026-11-11
- OpenClaw v2026.11.3 patch fixes cert verification and credential exposure across 56 PRs — dean_rie · 2026-11-11
- Fable 5.2 captured 1 in 4 Claude replies on day one, then growth stalled — dean_rie · 2026-11-11
- QA engineer proposes MCP agent to automate ad postback checks before every app release — Soobhee_Chukwutosin · 2026-11-11