3 commits

Author	SHA1	Message	Date
Sam & Claude	4192574f74	docs(hive): add 2026-06-20 status block — MVP code-complete, first-proof path ... Persist the reconstructed plan: all four MVP steps code-complete on colibri main; first proof is not code-blocked (interim manual runbook path); open work categorized (hardening #100/#92, CLI-driveability #101/#102, naming #98/iso#70). PR #90 (tenants table) closed as superseded — already on main. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-20 09:00:09 +02:00
Sam & Claude	7c3016cf06	docs(hive): record first-proof policy — scratch jail + test collection until hardening ... First proven end-to-end uses a scratch jail + throwaway test collection only; no real tenant data until path hardening (#92) lands. First-proof blockers are #88 (resolve collection by name) and #89 (per-call unlock); #92 is hardening. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-20 06:39:31 +02:00
Sam & Claude	21a630cd93	docs(hive): add HIVE-ONBOARDING design — colibri-vault as the join-the-hive primitive ... Captures the strategy: promote the proven clawdie-vault-fetch flow into a colibri-vault crate that provisions a jail's secrets, making 'spawn jail -> vault-provision -> register' the single onboarding step. - tenant = jail = bucket (1:1:1); Org Collections for multi-tenant isolation - 'one key' ideal (customer provider key + operator org service-account) - mother skill (genesis + policy-gated self-replication); colibri-vault is a limb - moat = capability routing (one-key agent borrows the whole multi-OS swarm) - security invariant: bootstraps on host, jails hold only resolved secrets - lean MVP + explicit overengineering traps to avoid Cross-linked from CAPABILITY-ROUTING. Building blocks LIVE, platform PLANNED. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-19 21:03:50 +02:00