Open source · 2026 - contextq
Local queues for AI agents — named queues with built-in context, FIFO claims, and a durable journal on disk. No daemon, no server.
- Go
- CLI
- AI-assisted development
Modern CLIs answer to two readers: humans and agents. Once an agent is the one invoking your tool, the shape of its output, its error model, and the protocol it enforces all become first-class design surfaces.
contextq is one small experiment in that direction: a tiny local CLI for coordinating agent work. Named queues carry a required block of agent-facing context, work items are bare string keys, and item pop is the only way to take work — it atomically claims the oldest available item under a per-queue file lock. No daemon, no server, no network. Just files on disk and a small Go binary.
Narrow, predictable, deterministic
Good agent-facing tools give agents structured answers instead of ambiguous prose — keeping context windows focused, response times short, and behavior on a known path. contextq returns one shape every time: --json with stable error codes means the agent branches on a field, not a paragraph.
A small model, deliberately
Three nouns — queue, key, item lifecycle — and five states with five legal transitions. Everything is appended to a per-queue events.jsonl you can inspect with cat and replay to recover state. The CLI rejects races, duplicate work, and invalid state transitions at the boundary, so an agent’s plan can’t drift into shapes the system was never built to handle.