Terminal Night 3 report: UNIX philosophy, CLI organization, and AI-era terminal design
4310sy · x · 2026-10-17
A Zenn writeup of Japan's Terminal Night 3 event, covering several substantive sessions:
- UNIX thinking (Songmu): multi-purpose programs are bad; interactive programs can't be piped and block on human input; write filter-style programs, send unrelated info to stderr; "silence is golden" — no stdout on success; keep colors/emoji filterable; use JSONL I/O for agent development; prefer man pages over mysterious --help output.
- CLI organization (Shisato): manage bookmarks with buku + fzf, control browser tabs via tabctl, pair Claude Code with nb to auto-summarize bookmark notes.
- AI-era terminal design (yasunori): AI speeds up building but makes operations harder — design loosely coupled terminals around the UNIX principle of doing one thing well.
More from coding & agent
- One user made 24,000 Copilot requests in 24h: how firms audit 580 AI seats nobody tracks — alex_verem · 2026-10-17
- Sports odds MCP server compares Pinnacle fair odds with Polymarket and Kalshi prices — Unhappy_Composer_894 · 2026-10-17
- Dev adds MCP server to a dance directory, shares Claude vs ChatGPT vs Grok testing notes — Ajunjahi · 2026-10-17
- Palo Alto Networks ships Cortex XCOR: AI SRE agent diagnoses alerts in under 3 minutes — shashib · 2026-10-17
- HF dev begs OpenAI/Anthropic to fix agents picking the wrong Chrome window — NielsRogge · 2026-10-17
- An agent negotiated a 3PL contract in 3 days, end to end — aarthir · 2026-10-17