cat architecture.md

MINT runs a local-first research infrastructure on a Mac Studio. Scheduled services collect research material, maintain a searchable paper corpus, prepare a weekday news briefing, and support lab work in Slack. Public pages are static files deployed through GitHub Pages.

trace content-pipeline

FeedMe is the shared collection and ranking layer. The full 04:00 run invokes each source adapter, removes material seen in earlier runs, classifies every new item, summarizes eligible longform pieces, ranks the results, and builds a story-aware curator slate. The 12:00 and 20:00 runs refresh X, Bluesky, and arXiv without repeating the full source sweep.

Source adapters

Products and agents

The Morning Briefing product was retired on 11 July 2026, and current FeedMe runs do not invoke it. Retained modules supply reusable full-text retrieval and fallback-ranking code. Daily #feed-* posting was retired on 18 July; selector auto-promotion to #firehose remains active.

trace corpus-ingestion

Paper ingestion is a queue-backed set of focused services. Local folders and Slack provide intake, the handoff service coordinates work, analysis prepares structured outputs, one writer owns database mutation, and a final tidy stage handles downstream organization.

1. corpus-intake-localWatch local paper intake.
2. corpus-intake-slackReceive papers and links from Slack.
3. corpus-ingest-handoffValidate and dispatch queued work.
4. corpus-analyzeConvert, embed, enrich, and analyze papers.
5. corpus-db-writerSerialize durable LanceDB writes.
6. corpus-tidyFinalize files, catalog records, and reports.

du -sh corpus/

Each paper is represented at several levels: bibliographic metadata, full text, searchable chunks, embeddings, structured research-question analysis, and hierarchical summaries. The current paper embedding policy is Voyage voyage-context-3, 2048d.

Explore the public projection in the Corpus Map. The public map exposes research structure and links, while the working corpus remains on lab infrastructure.

query @Minty

The corpus agent combines lexical and vector retrieval over papers, chunks, semantic dimensions, and summaries. It supports quick lookup, broad literature review, similarity search, and catch-up requests. Results are grounded in local corpus records and returned with source links.

cat AGENTS.md

Minty is the lab's research-management agent. The root AGENTS.md is the always-on contract for identity, operating boundaries, verification, and collaboration. Project-specific THREAD.md, HANDOFF.md, LOG.md, and TODO.md files carry workstream context.

The old root fact digest and session database are retired archives. Durable lessons belong in the active runtime's memory, and exact history stays in that runtime's native transcript store.

launchctl list | sort

launchd owns persistent and scheduled services. The table below is generated from Minty's canonical daemon registry and reconciled with the jobs currently loaded on the host. 45 user jobs and 2 system jobs are loaded.

calendar daily-jobs

Exact launch intervals and loaded state are listed above. The schedule table highlights the main lab-facing rhythm rather than duplicating every watchdog and maintenance interval.

help lab-members

Bots

Slack channels

inspect agent-runtime

Scheduled agents run through a shared Codex router with explicit model and reasoning policy. Deterministic services use ordinary code where a model adds no value; research synthesis and judgment-heavy work use higher reasoning effort. Daemon subprocesses receive explicit virtual environments and isolated runtime homes.

ls integrations/