Observed repository visibility: PUBLIC.
Public, one-repository research infrastructure for human-directed collaboration across ChatGPT, Claude, local tools, GitHub, terminal workflows, and future interfaces.
➡️ Begin with README_START_HERE.md.
Articles, tweets, thoughts, plans, sketches — public by design, polish optional:
board/ · firehose board/INBOX.md
Parallel thrash branches with auto CI; main stays clean:
lab/README.md · registry lab/BRANCHES.md
Abuse worktree: /home/anon/Lab-PLAY · branch play/operator-abuse-sandbox
Press notes, tutorials, and human-facing statements: communications/.
Research-readiness statement (human judgment; not a product cert):
communications/statements/2026-07-22_READY_FOR_SERIOUS_RESEARCH.md.
Load-bearing. Read WHY_NOT_TO_TRUST_THIS_PROJECT.md before treating anything here as safe, authoritative, or product-ready. It lists permanent reasons rooted in anonymous non-coder operation, AI seats, research-only doctrine, and structural limits that do not go away when tests pass.
one repository = one authoritative corpus namespace
Git branches = proposal spaces
main = accepted working state
Git tags = exact historical anchors
snapshots = deterministic CANONICAL_AS_IS artifacts
audit overlays = append-only observations bound to target hashes
AI accounts = replaceable, attributed research seats
Repository visibility was resolved as PUBLIC by
DEC-2026-000002, and the
constitutional privacy model was reconciled accordingly. Public visibility does
not waive LOCAL_ONLY, SECRET, provider-routing, or redaction rules.
R001 implements a working operator shell, deterministic snapshot creation and verification, provider-neutral route packs, a read-only Claude audit pack, hash-chained observation ingestion, GitHub Actions checks, privacy/secret scanning, and one-command GitHub bootstrap logic.
Read docs/DEMONSTRATION.md for the claim-to-artifact map.