Context sovereignty: treat every agent input as a stranger
gerardsans · x · 2026-11-05
Continuing his agent-engineering series, the author covers team practices for keeping humans in control: keep specs and rules as plain files in your repo rather than trapped in vendor chat history (context sovereignty), with a promotion path from personal tokens to sandboxed work to budgeted production repos; and treat every external input as a stranger—models can't reliably tell ephemeral context from orders, so scrub and quarantine everything before the agent sees it.
Related event: Agent Team Engineering: Context Sovereignty and Tiered Autonomy(2 posts)→
More from coding & agent
- Auto-unsubscribe agent built on Jev: browser-aware email triage cleans inbox in a week — threepointone · 2026-11-05
- Olki ditches LLM-based agent monitoring for strict type systems and effects algebra — bigdata · 2026-11-05
- Alpha School Launches Founders School: Graduate With a Business Doing $1M — nateliason · 2026-11-05
- Why Coda's formula language beats Notion for deterministic agent memory — bfrench · 2026-11-05
- Dev quits 11-year career, vibe-codes AAA-looking sci-fi FPS with AI in 24h, lands Gamescom booth — Aiden_Tech_Ai · 2026-11-05
- JetBrains argues IDEs still matter in the agentic coding era: 'a cockpit, not a typing tool' — Sam_Witteveen · 2026-11-05