
Our Brand Guidelines are public. They fix the tagline, voice (operational, understated, honest), project spelling, and a 'terms to avoid' list covering overused industry phrases. Same reference for every project. https://nanocollective.org



No application form. Maintainers at the Nano Collective are added by invitation: consistent contribution, judgement that fits, willingness to take on the work. Stepping back is welcome.

https://nanocollective.org



`get-md`'s batch API runs an async iterator with bounded concurrency, exponential-backoff retries, and an opt-in file-system cache. Re-runs while you tweak chunk size or LLM config finish near-instantly. https://github.com/Nano-Collective/get-md



json-up's migrate() throws three named errors you can branch on: ValidationError (schema mismatch, exposes Zod issues), MigrationError (up() threw, exposes the cause), VersionError (config bug). No generic catch needed.

https://github.com/Nano-Collective/json-up



Shift+Tab into Plan mode in Nanocoder. The model gets read-only tools, investigates your codebase, and writes a structured plan: files, steps, risks, open questions. No file writes or shell until you switch back to execute. https://github.com/Nano-Collective/nanocoder



`nanotune benchmark` runs your tests and writes a JSON + markdown report: pass rate, latency, TTFT, tokens/sec, per-test results. Read the failures, add targeted examples with `nanotune data add`, retrain, re-run. https://github.com/Nano-Collective/nanotune
