brain/
← all sparks
sparkgraduatedwarm2026-06-07voice→ /queue

Finish the research queue system

Finish the research queue system

The spark

I would like to finish building the Queue system that captures research ideas and queues them up.

What I think you're reaching for

The capture→act loop is half-built and you want to close it. The schema exists (QUEUE defines the entry shape and lifecycle) but the queue is empty and there's no friction-free way to add to it or drain it. "Finish" means the two open ends: an easy on-ramp (a /queue capture verb, or sparks graduating straight into a queue entry) and an automated off-ramp (pending entries getting picked up by /autoresearch / /academic-research without you hand-shepherding each one).

Connections (held loosely)

  • QUEUE — this is the thing to finish, not start. The YAML shape, statuses, and lifecycle are already specified; what's missing is the tooling around it. Worth re-reading it before building so you build to the schema you already designed.
  • This is the off-ramp the 2026-06-07-brain-as-add-capture-layer spark implies. Sparks catch the idea; the queue catches the research intent. A ripe spark's natural graduation target is a queue entry — they're two stages of one pipeline (capture → queue → research → wiki), and the spark skill already names "queue a _meta/QUEUE.md entry" as a next move.
  • The research skills (/autoresearch, /academic-research) are the queue's consumers — finishing the queue means wiring a "pick the next pending entry and run it" path, and QUEUE.md already anticipates a web app management UI in apps/web as the eventual front-end.

Where this could go

  • A /queue skill — the missing capture verb: /queue <topic> appends a well-formed entry (slug, method, context, priority) so the on-ramp is as frictionless as /spark.
  • Spark → queue graduation — make "graduate this spark" able to target a queue entry directly, closing the capture→research seam end to end.
  • Drain automation — a step (DAILY-style, or on demand) that takes the top pending entry, runs the right research skill, and flips it to in-progress, so the queue actually moves instead of accumulating.

A possible next move

This one's a build task, not a marinate — the cleanest move is a _meta/QUEUE.md entry or a scoped plan for the /queue skill + drain step. If you want to start now, say so and we can scope it; otherwise it sits here as the reminder that the loop is half-closed.