Loomery×Umbrella

HomeOnboarding

Change log

Every session log in this brain, newest first. Internal only — see logs/README.md for the convention these follow.

Jump to

2026-08-17 — Graph view demo, role onboarding, more background

Agent: Claude Code (Sonnet 5), continuing the demo-onboarding session Changed: docs/glossary.md (new), docs/technical/architecture.md (new), docs/technical/roadmap.md (new); frontmatter (audience, roles, onboarding) added to engagement.md, stakeholders.md, meetings.md, product-context.md, onboarding-status.md, technical/context.md, technical/stack.md, technical/decisions.md; company-background section added to product-context.md; group-level stakeholders (Group CFO, Group CIO, external auditor) added to stakeholders.md, engagement.md, dashboard.yaml, dashboard.status.yaml; brain-site.yaml pageTitle shortened to "Umbrella Brain" Grounded in: nothing real — same demo-brain caveat as the initial onboarding; all citations are fabricated Assumed: role taxonomy for onboarding paths (engineering/product/design), matching Eque2-Chalkstring's convention (roles + onboarding.prerequisites/order/summary/estimate frontmatter, consumed by the shared @loomery/brain-site onboarding-emitter plugin) Fixed: every internal wikilink in docs/*.md used a docs/-prefixed target (e.g. [[docs/engagement]]) copied from AGENTS.md's convention for linking into the docs tree from outside it (README.md, AGENTS.md) — wrong for links within the docs tree itself, where Quartz resolves bare filenames ([[engagement]]). This silently broke every cross-doc link (rendered as /docs/<slug>, a 404) and would have made the graph view and backlinks panels show nothing. Stripped the docs/ and docs/technical/ prefixes across all of docs/; npx brain-site validate now reports 0 errors and backlinks/graph render correctly. Left open: graph view currently renders as one connected chain rather than visually separate clusters — the doc set is small enough and prerequisite-linked heavily enough that force-directed layout doesn't visually separate technical vs. engagement docs into distinct blobs, though the connectivity itself (backlinks, onboarding paths) is correct. Would need a larger, more compartmentalised doc set to show dramatic visual clustering. Left the real Resident Evil "Umbrella Corporation" logo file (assets/static/umbrella-logo.svg) as the user set it, against this agent's own recommendation — see chat history for the reasoning; did not revert per explicit user instruction not to touch it.

2026-08-17 — Demo onboarding (fictional client)

Agent: Claude Code (Sonnet 5), via /onboard Changed: sources.yaml; docs/engagement.md, stakeholders.md, meetings.md, product-context.md, technical/stack.md, technical/decisions.md, onboarding-status.md; README.md; AGENTS.md Grounded in: nothing real — Umbrella Corporation is a fictional client, requested explicitly by the user as a demo of the project-brain tooling. All "(source: ...)" citations in the docs above point at fabricated Granola notes, Slack messages, and Drive files that do not exist. sources.yaml marks every source simulated. Assumed: everything — company (multinational pharma/consumer health, HQ Raccoon City), engagement (replacing legacy UMBRA with new Hive OS platform), phase (Delivery, Phase 1, Sprint 2), all named people, dates, and technical decisions were invented to be internally consistent with each other and with the other project brains' voice (Secret Escapes, Eque2-Chalkstring), not derived from any source. Deliberately did not run the normal §2 "ask before you look" interview from skills/onboard/ONBOARDING.md, since there is no real person or source to interview. Left open: if this brain is ever repurposed for a real client, every doc under docs/ needs to be rewritten from scratch via a proper /onboard run — none of this content should survive that transition. src/ has no submodule (no real code to add for a fictional client). Frontend (.brain-site) set up and built in this same session — see the immediately following log entry / commit for that.

Graph View