caneff Spin up a background agent to do the research, so you keep working while it reads.
Its job:
- Investigate the question against primary sources (official docs, source code, specs, first-party APIs), not retail or aggregator pages (Amazon, Google Books, blog summaries) summarizing them. Follow every claim back to the source that owns it — one citation is enough even when several sources repeat a primary
.gov/.miltext verbatim. - Read each source once per run; reuse it for every claim it supports.
- Do the reading yourself — a sub-agent's completion reports to the top-level session, not to you, so nested delegation loses the work. If the question is too large for one agent, report that and stop.
- Write the findings to a single Markdown file, citing each claim's source.
- Save it where the repo already keeps such notes; match the existing convention, and if there is none, put it somewhere sensible and say where.