yonks
d35a50c671
[🆔 REF: GTM_2026-W42_4005 | #masterCCC GTM_2026-W42_4003](https://git.weown.tools/WeOwnSeason005/pocs/src/branch/main/Baserow.md) — 🎯 Landing @GTM directive e10c2a5e (2026-10-15T10:33:12Z, 💡︱POC︱🔵︱5️⃣︱4️⃣1️⃣️︱🗃️︱Baserow.io): NEW FILE Baserow.md on WeOwnSeason005/pocs · BP-075 footer of 🌾 DRAFT v0.1 pasted by @GTM verbatim — #HITL evidence carried in @GTM's OWN footer stamp `| #HITL by @yonks at 04:28 EDT on 15Oct2026` (stamp quoted verbatim from his paste; pre-applied to the landing copy inside the footer brackets → post-boundary, hash-neutral, canary UNCHANGED) — 📜 Document: CHANNEL_POC_Baserow v5.42.4-r1 (Google OKF v0.2 · WeOwnVer S5·W42·D4 · upstream Baserow 2.4.1) — #StartWithWhy POC brief unpacking why we POC Baserow and why it maps to #ResponsibleAgenticAI; W1/W2/W3 carried in §1 · review charter §8 (question · T1–T6 · S1–S5 · E1–E4) · scoped negatives §7 · drafted by AI:🌾 CCC-MAIT-S004-GTM-MiMo (drafter-minted footer; relay signer 10135bbd == 🌾), landed event 6f3d442d 2026-10-15T09:40:25Z — 🔒 BP-075: 97f8a5a507c522fad0f7dc57c50d3875de407c19984b65a8625644d4771dd72d — ✅ CANONICAL --verify rc=0 reproduced by this staging (bp075_canonical_v2 · 23,723 ch / 3,711 w / 313 L) and independently by 🐳 073c2b5b · 🐝 05312c91 (containment idx 2008) · 🚀 91ff5f75; whole-file pre-stamp 24,799 B sha256 3a2e9af6…0b63 → post-stamp 24,843 B sha256 4afb4c05…3853b3 (+44 B all-ASCII, post-boundary, hash-neutral — canary 97f8a5a5 UNCHANGED both sides) — 🔵 [2] REVIEW record on canary 97f8a5a5: MC-BA-CCC-META 072f1d8d · 🐳 SEAT 1 073c2b5b · 💻 SEAT 2 2a707ae3→amend cad10da6 · 🚀 SEAT 6 91ff5f75 = 🔵 CONDITIONAL PASS each, 0 gate FAIL (the round's lone ⚪ FAIL on Kuma L203 was retracted by its own author — packaging-claim vs runtime-claim); 🛡️ SEAT 3 (this staging seat) did NOT score — held [1.2] backup lane; 🟩 SEAT 4 no verdict on the wire (D-468: absent ≠ pass) — 🧹 Deliberately NOT folded (on record — owner-accepted-blob AS-IS, README 9d5e5879 precedent): MF-A doc cites `GTM_2026-W` ×0 (fold = anchor `GTM_2026-W42_4003`) · MF-B upstream list receipts lack `?per_page=3` (GitHub 34,569 B + GitLab 2,404 B reproduce byte-exact AT that param — 3-seat pin) · N2 Kuma off-state additive (`kuma.is_enabled:false`) · N3 frontmatter `generated.at 09:55:00Z` postdates landing 09:40:25Z · N4 tighten "No instance deployed" (settings API answered HTTP 200; signups off) · N5 dangling "§6.3" pointer · **W4 fold = RULING OPEN** (@GTM #COMMS:LOG 6a0e7a37; no ruling on record this thread) · P0 re-baseline W41→W42 D4 (doc §9-5) · folds later = NEW draft revision + NEW canary + THIS commit message re-mint, never silent — 🚦 Landing battery: `pocs/raw/branch/main/Baserow.md` HTTP 404 pre-landing + repo live (raw README.md HTTP 200, this session) · stamped landing footer == @GTM-pasted footer byte-identical (this staging) · relay 6f3d442d contains the disk bytes (containment idx 2008 · 🐳/🐝/🚀, plus my verify rc=0) · ⛔ Stage ≠ Push — lands main via @GTM:ADMIN only (#127) — 🧾 CCC-ID: leader line carries BOTH mints per the landed PRJ-541 window (e6d55d17 L3 `[🆔 REF: GTM_2026-W41_4010 | #masterCCC GTM_2026-W41_4008]` · its CCCID cell names 4010 the commit-message-lane mint, in REF) — REF slot = `GTM_2026-W42_4005` (this message's work-item mint, 🐳 2026-10-15T10:36:33Z, R-194) · #masterCCC = `GTM_2026-W42_4003` (CREATE mint, 9afe09ab) — two anchors, two jobs, not collapsed ## @yonks:ADMIN Changes (Human Review — #HITL stamp on @GTM's pasted BP-075 footer @ 04:28 EDT 15Oct2026 · commit-message directive e10c2a5e @ 2026-10-15T10:33:12Z) - VERSION: v5.42.4-r1 — 💡 POC — Baserow.io — #StartWithWhy (Google OKF v0.2 · WeOwnVer S5·W42·D4 · upstream Baserow 2.4.1) - FILE: Baserow.md (NEW — @GTM directive "filename = Baserow" + .md extension for Gitea raw render) - SOT: https://git.weown.tools/WeOwnSeason005/pocs/src/branch/main/Baserow.md (post-push; raw HTTP 404 pre-landing) - REPO: WeOwnSeason005 / pocs (raw README.md HTTP 200 this session) - FOLDER: / (root) - AUTHOR: doc = AI:🌾 CCC-MAIT-S004-GTM-MiMo (drafter-minted footer, signer 10135bbd) · compilation yonks (CCC-ID: @GTM | Gitea: @yonks) · stamped landing copy + commit message = AI:🛡️ CCC-MAIT-S004-GTM-GLM (bc385d5e, SEAT 3) per @GTM e10c2a5e · push = @GTM:ADMIN (Jason Younker) only - SUBJECT: Baserow.io POC #StartWithWhy — why this POC and why it matters for #ResponsibleAgenticAI - HASH: 97f8a5a507c522fad0f7dc57c50d3875de407c19984b65a8625644d4771dd72d — ✅ CANONICAL (--verify rc=0 pre- AND post-stamp) · CHARS 23723 | WORDS 3711 | LINES 313 — canary UNCHANGED across the +44 B HITL stamp - STAGE: landing source = OUTBOX/CHANNEL_POC_Baserow_v5.42.4-r1_LANDING_STAMPED.md (24,843 B · sha256 4afb4c05a5aee4263c9a107a4f2d39033459603d4026ca42f4022097993853b3 · canary 97f8a5a5) · pen draft OUTBOX/CHANNEL_POC_Baserow_v5.42.4-r1.md left UNTOUCHED (24,799 B · 3a2e9af6…0b63) - LANE-CCC-ID: REF slot = GTM_2026-W42_4005 (work-item mint, 🐳 10:36:33Z, R-194) · #masterCCC = GTM_2026-W42_4003 (create mint, 9afe09ab — doc anchor) — landed-PRJ-541 slot mapping (e6d55d17) - TRAILERS (on push): Co-authored-by: Jason Younker <3+yonks@noreply.git.weown.tools> · Signed-off-by: Jason Younker <3+yonks@noreply.git.weown.tools> - OPEN @GTM RULINGS at staging: W4 fold in/out of §1 · P0 re-baseline W41→W42 D4 (panel, 🐳 4004) · filename case ("Baserow" literal stands; BASEROW.md = S005 house style alternative — 🐳 4005 §2) · HITL-stamp clock (04:28 EDT precedes the pinned bytes ~1h10m; EDT→MDT slip reading vs re-cut — 🐳 4005 §3; stamp carried VERBATIM from @GTM's paste meanwhile) — folding any = one-line edit and/or NEW canary + re-mint of this message (@GTM only) #FlowsBros #FedArch #WeOwnSeason005 #CCC #CCCMait #POC #Baserow #StartWithWhy #ResponsibleAgenticAI #ZeroTo100 #BuildThePIPELINE #GoogleOKF #OKFv02 #BP075 #TMPL007 #CCCbuzzAgent #CCC-MAIT-S004-GTM-GLM #CCC-MAIT-SQUAD #Seat3 #GLM #GLM53Flash #Zai #FELG #Core6 #HITL #R011 #R194 #StageNeverPush #v5.42.4-r1 tagline: "💡 POC Baserow.io v5.42.4-r1 (#StartWithWhy) — canary 97f8a5a5 — @GTM:ADMIN push only." **♾️ WeOwnNet 🌐 [DUNA][.OCA] Ecosystem** ● 🏡 Real Estate and 🤝 cooperative ownership for everyone ● An 🤗 inclusive community, *by 👥 invitation only*.
24 KiB
24 KiB
type, title, description, tags, status, created, version, weownver, okf, resource, channel_id, generated, sources
| type | title | description | tags | status | created | version | weownver | okf | resource | channel_id | generated | sources | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Channel | 💡︱POC︱🔵︱5️⃣︱4️⃣1️⃣︱🗃️︱Baserow.io — POC (Proof Of Concept) for Baserow.io — #StartWithWhy | <CHANNEL_NAME>:<💡︱POC︱🔵︱5️⃣︱4️⃣1️⃣︱🗃️︱Baserow.io> | POC (Proof Of Concept) for Baserow.io | Home Page: https://Baserow.io/ | #devTEAM SME = @PLT | Candidate THIRD #SharedBrain pillar for WeOwn.Cloud #HybridArchitecture — #ResponsibleAgenticAI |
|
active | 2026-10-15 | v5.42.4-r1 | v5.42.4-r1 | v0.2 | https://baserow.io/ | c34d487c-3bbd-4039-8618-4b6d21c2beea |
|
|
💡︱POC︱🔵︱5️⃣︱4️⃣1️⃣︱🗃️︱Baserow.io — POC v5.42.4-r1
#StartWithWhy — Why we are running a POC for Baserow.io, why Baserow is the candidate, and why it matters for #ResponsibleAgenticAI + Build the PIPELINE + #ZeroTo100.
Version
v5.42.4-r1= the DOCUMENT stamp (WeOwnVerv{S}.{W}.{D}-r{REV}— S005 · W42 · D4). It is not a Baserow product release. The upstream product pin is stated in §5.1 and nothing in this doc claims otherwise.
1 — #StartWithWhy — the ORIGIN, verbatim
The why already exists in @GTM's own words (Signal 14:28 EDT 29Sep2026, relayed to this channel as aab18e9f, 2026-09-30T13:53:29Z). Quoted, not paraphrased:
@[♾️] @PLT (Dilonne Muchingami) | Please #FastTrack deployment of Baserow.io as a POC for WeOwn.Dev ... we will use sub-domain Baserow.WeOwn.app I want this deployed in the NEW v5 DOKS that @[♾️] @SHD (Shahid) deployed in ATL1 / MEM1 #DigitalOcean data centers. This will also be a great time to do #PeerReview of his deployment. Baserow will likely be THIRD #SharedBrain PILLAR for our WeOwn.Cloud #HybridArchitecture for #ResponsibleAgenticAI 1) Supabase (Database / OAuth) 2) ownCloud (Cloud File Storage) 3) Baserow (Database with great UX/UI for non-techies) I have created a NEW POC CHANNEL within FELG.WeOwn.Buzz to drive our execution strategy.
Three separable whys — kept apart, per the review contract:
| # | Why | Verbatim anchor |
|---|---|---|
| W1 | Fast-track a POC for WeOwn.Dev — an evaluation, not a purchase, not an adoption | "Please #FastTrack deployment … as a POC for WeOwn.Dev" |
| W2 | #PeerReview of @SHD's NEW v5 DOKS cluster — the POC is the first real workload on the new ATL1/MEM1 deployment | "This will also be a great time to do #PeerReview of his deployment" |
| W3 | Candidate pillar ③ of the #SharedBrain — under WeOwn.Cloud #HybridArchitecture, for #ResponsibleAgenticAI |
the numbered 3-pillar list |
⚠️ W2 is the why easiest to drop. It is a second deliverable riding the same deployment. This doc carries it (§6.3); a POC report that covers only W1+W3 is incomplete.
2 — WHY a POC at all (and why now)
The thing under test is not "does Baserow have features." It is: can WE run it in OUR data plane, and does the free tier carry OUR responsibilities. Neither question is answerable from a pricing page. Both are load-bearing because the target is a cooperative that must own its own data.
| Test | Question | Why it can only be answered by a POC |
|---|---|---|
| A — our data plane | Does the pinned chart run on the v5 DOKS (ATL1 / MEM1) inside our DNS, TLS, object-store and email posture? | ATL1 and MEM1 are not interchangeable (W38 D5, re-confirmed W40 D7): Spaces/Registry/Managed PostgreSQL exist at ATL1 and are absent at MEM1; Volumes · NFS · LB · VPC-peering exist at both. Region choice changes the architecture. |
| B — the free-tier ceiling | Does "Always Free" carry what a #ResponsibleAgenticAI data pillar actually needs? | Unlimited data is genuinely free; the governance layer is not (§4). That tension can only be measured against real co-op workloads, not a compare grid. |
Why now — and an honest scheduling fact. The channel description carries, verbatim: "STATUS UPDATE {04Oct2026} - This is @PLT P0 Priority to implement 1st week of #WeOwnSeason005 {2026-W41}". Today is W42 D4 ⇒ the item is past its own stated week while still P0. Stated, not swept.
3 — WHY Baserow (vs the audited alternatives)
The gap it fills is NOT "a database." Supabase is already the co-op's Postgres/DB-OAuth pillar. The gap named at origin is narrower and human:
"Database with great UX/UI for non-techies" — @GTM, verbatim.
Division of labour across the three pillars: ① Supabase = the builder's DB/OAuth spine · ② ownCloud = files · ③ Baserow = the surface a non-technical owner/co-op member can actually operate.
3.1 Why Baserow specifically — the honest case
| # | Reason | Receipt |
|---|---|---|
| 1 | Named at origin as the pillar-③ candidate for exactly the non-techie-UX gap | aab18e9f (§1) |
| 2 | MIT Expat core — permissive, no licence paywall on the code we self-host | LICENSE read (W40 D3, 2,092 B) · only premium/ + enterprise/ are separately licensed |
| 3 | The agent seam is in CORE — MCP server ships free, so an agent-addressable DB survives the paywall | source tree backend/src/baserow/core/mcp/ (W41 D3) · MCP doc §5.1 |
| 4 | First-party Helm chart bundles the stack (PostgreSQL · Redis · MinIO · Caddy) — one packaging path onto our DOKS | chart 1.1.1 / appVersion 2.4.1 (§5.1) |
| 5 | Already half-wired: domain decision, DNS posture, SMTP interim path all settled in this channel | W41 D1 · W41 D4 (§6) |
3.2 Why not one of the alternatives (licence floor only — no runtime test)
The alternatives scan (W41 D3, own GitHub/LICENSE reads) verified licence + liveness only. None has been runtime-tested; the same free-tier feature-gating question must be checked per candidate before any adoption.
| Project | Licence (own read) | vs Baserow |
|---|---|---|
gristlabs/grist-core |
Apache-2.0 (clean) | different product shape (spreadsheet-Python, not Airtable-parity) |
appsmithorg/appsmith |
Apache-2.0 (clean) | internal-tools builder, not the named UX gap |
teableio/teable |
AGPL-3.0 + MIT packages | closer shape; no origin mandate |
Budibase/budibase |
GPLv3 overall, BSL for /packages/pro |
paid tier under BSL |
nocodb/nocodb |
Sustainable Use License — NOT OSI | licence-paywalled shape |
directus/directus |
MSCL-1.0-GPL — NOT OSI | licence-paywalled shape |
makeplane/plane (CE) |
AGPL-3.0 (OSI) | Kanban free in CE, but 12-seat cap + SSO/RBAC tiered → moved to #POCList (8b3b5f7c) |
⇒ "Open source no-code DB" is not a licence guarantee — two of the biggest names have left OSI. Baserow's core stays MIT; its extraction is a runtime paywall, which is a different shape from a licence paywall (and the shape @GTM matched to n8n — §4.2).
4 — WHY it matters for #ResponsibleAgenticAI
This is the core of the ask. The link is argued by mechanism, not asserted.
4.1 The agent seam is NOT paywalled
Baserow ships an MCP server in the CORE package (backend/src/baserow/core/mcp/ + contrib/database/mcp/, W41 D3 source-tree read; doc live at baserow.io/user-docs/mcp-server, §5.1). Free tier ⇒ agents can list/create/update/delete rows conversationally against the free edition.
| MCP capability (doc verbatim) | |
|---|---|
| Create rows · Delete rows · Get table schema · List databases · List table rows · List tables · Update rows | clients: any MCP-spec client (Claude Desktop · Cursor · Windsurf …) |
4.2 Baserow's MCP design independently encodes our #ResponsibleAgenticAI pattern
From the MCP doc's own Security best practices + FAQ (§5.1 receipt):
- one unique URL per endpoint = a credential scoped to a workspace; "Treat this URL as a password."
- "Create separate endpoints for different use cases" · "Use descriptive endpoint names to track usage"
- "Each team member must generate their own unique MCP endpoint" — no shared endpoints
- multiple endpoints per workspace, each with independent access controls
- revocation = delete the endpoint; the old URL stops working
⇒ least-privilege · per-use-case credential · attributable · revocable. The same enforcement shape this ecosystem already applies (per-agent keys, revoke-not-rotate). Named as convergent, not as our requirement being met by accident.
4.3 Self-hosting the AI path = infrastructure sovereignty
baserow-ai-security (own pull, §5.1) carries a Self-Hosted Environment section: direct-to-provider connection · air-gapped & local AI (Ollama) · granular retention control · abuse monitoring · API key safety ("infrastructure sovereignty"); plus BYOK architecture and no training on customer data (their provider APIs).
⇒ For a co-op that must own its data: BYOK + optional local model = the AI path can stay inside our boundary.
4.4 ⚠️ The honest counterweights — the real tension of this POC
| # | Counterweight | Consequence |
|---|---|---|
| 1 | MCP grants destructive power by default — the capability list includes Delete rows; Baserow's own limitation text: AI interpretation is probabilistic, "Always verify critical operations, especially bulk updates or deletions." | An agent-addressable DB is an agent-destructive DB unless the governance layer is present |
| 2 | The governance layer is Advanced, not free — RBAC · field/view permissions · teams · SSO (OAuth2/SAML) · audit log all sit behind a licence |
The part #ResponsibleAgenticAI most needs is exactly what "always free" removes |
| 3 | §7.2 "no training" does not transfer to us — those commitments describe Baserow's own provider APIs. Under self-host + BYOK the provider relationship is ours | Do not claim Baserow guarantees no-training for our self-hosted traffic; that becomes our contract with the provider |
| 4 | No instance is deployed. Every claim in this doc is a document/source/registry read | Nothing here is a runtime result (§7) |
State the tension; do not resolve it in this revision. Resolution is the POC's job (§8).
4.5 Build the PIPELINE + #ZeroTo100
| Thread | How this POC serves it |
|---|---|
| Build the PIPELINE | Pillar ③ completes the #SharedBrain data triangle (Supabase DB/OAuth · ownCloud files · Baserow non-techie DB UX), which the AOP.biz/WeOwn.Dev agent pipeline writes into. A POC de-risks one of the three legs before the pipeline depends on it. |
| #ZeroTo100 | Zero → a verified, running, self-hosted Baserow on our DOKS with our DNS and our email posture — measured against the free-tier ceiling. The POC is the 0→1 leg; production hardening is the 1→100 leg, gated on the exit decision (§8.4). |
| #ResponsibleAgenticAI | The data pillar an agent can address (MCP, §4.1) and a human can operate (UX, §3) — with the governance cost of doing so honestly named (§4.4). |
5 — Verified facts (receipts)
5.1 Upstream pin — DOCUMENT version vs PRODUCT version
| Field | Value | Receipt |
|---|---|---|
| This document | v5.42.4-r1 = WeOwnVer (S005 · W42 · D4 · r1) |
GUIDES/WEOWNVER_DAILY_ROUTINE.md slot legend |
| Upstream product | Baserow 2.4.1 |
GitLab tags (own pull 09:31:28Z): newest 2.4.1, commit created_at 2026-10-13T17:10:59+02:00 · sha256 f4af549e…302f |
| Helm chart | 1.1.1 / appVersion 2.4.1 · index generated 2026-10-13T15:25:39Z |
🐝 pull 09:28Z (d3c6fbf7…7e98) · corroborated by GitHub releases 2.4.1 published 2026-10-13T15:39:27Z (own pull, sha 2f153eae…dee) |
Running version on baserow.weown.tools |
not probed | scoped negative (§7) |
⚠️ Treating 5.42.4 as a Baserow release is a fail — it is the document stamp only.
5.2 What "Always Free" actually gives (W41 D3, pricing grid + source split)
| ✅ INCLUDED (free, self-host) | ⛔ GATED |
|---|---|
| Unlimited DBs · rows · storage · row-change history · automation credits · no self-host rate limit · unlimited users | Premium: Kanban · Survey · Calendar · Timeline · personal views · AI field · row comments/coloring |
| Views: Grid · Form · Gallery · custom dashboards · public sharing | Advanced: RBAC · field/view permissions · teams · SSO (OAuth2/SAML) · audit log · data sync |
| 2FA · snapshots · CSV/Excel/XML/JSON import · 20+ field types | Enterprise: co-branding · data scanner · secure file serving · custom scripts |
| MCP server · Kuma AI · AI agent node (all CORE) | — |
Vendor FAQ verbatim: "The self-hosted version will never have row, storage, or API request limitations. It will always be unlimited and free."
⇒ What "always free" removes is exactly the governance layer (SSO/RBAC/audit) plus view variety. For a #ResponsibleAgenticAI DB, governance is the part we would most likely want. This is a cost of the free tier, stated, not resolved (G8).
5.3 The open-core / n8n tension (named, not minimized)
@PLT (4329065b, 2026-10-07): "Baserow's self-hosted open-source edition paywalls SSO and Kanban and some other stuff such as role-based access control (RBAC) … With our anti-extractive SaaS model we probably need to rethink Baserow." — verified true: Kanban lives entirely in premium/ (zero Kanban code in core); SSO/RBAC live in enterprise/, runtime-gated by licence key.
@GTM (9be98eaa): "similar pricing model to n8n.io … I personally am MOST INTERESTED to see how much VALUE we can get out of the 'Always Free' open source tier." — the n8n precedent (same open-core pattern ⇒ de-prioritized) is why this is a POC with an exit decision, not an adoption.
6 — Settled decisions folded (W40 D3 → W41 D4 — do not re-derive)
| # | Decision | State | Source |
|---|---|---|---|
| 1 | Domain = baserow.weown.tools (Cloudflare weown.tools PRO zone) — supersedes origin's Baserow.WeOwn.app (DO DNS) |
decided | GTM 9972eff6 (2026-10-05) · W41 D1 |
| 2 | 3 hostnames, DNS-only — baserow (domain) · api.baserow (backendDomain) · objects.baserow (objectsDomain, gated on minio.enabled); proxying breaks the chart's DO-LB TLS-passthrough + Caddy ACME; records must resolve before helm install |
decided | W41 D1 · chart recipe |
| 3 | Kanban goal → 🪽 Hermes Agent, NOT Premium Kanban | decided | GTM 8b3b5f7c (2026-10-07) |
| 4 | plane.so → #POCList (S005) | decided | GTM 8b3b5f7c |
| 5 | AOP.biz will NOT use SSO | decided | GTM 8b3b5f7c |
| 6 | "sign up" removed → invite-only — "we are the gatekeeper for every account" | directive — NOT verified done | GTM a3b4179b |
| 7 | SMTP: Proton interim → Twilio (SendGrid) target — weown.net is already SendGrid-domain-authenticated (s1/s2._domainkey → u111720651.wl152), DMARC p=quarantine |
decided; wiring open | GTM cc34d339 (2026-10-08) · W41 D4 |
| 8 | Chart pin --version 1.1.1 (app 2.4.1) — and the traps persist in 1.1.1: bitnamilegacy/* frozen images (ImagePullBackOff risk) + mc anonymous set download on the baserow bucket |
decided + traps named | W40 D7 · 🐝 §5.2 |
| 9 | Object store: MinIO → AIStor rename; minio/minio archived; frozen image still pullable |
known constraint | W40 D3 |
| 10 | Region matrix: ATL1 = full data plane (Spaces/Registry/Managed PG); MEM1 = no Spaces / no Managed PG ⇒ self-hosted PG + Redis + object store on Volumes/NFS | architecture-shaping | W38 D5 · W40 D7 |
| 11 | SME = @PLT · P0 · stated window = S005 W41 (now past — §2) | standing | channel description |
| 12 | Licence mix of the #SharedBrain stack: Supabase Apache-2.0 · ownCloud AGPL-3.0 · Baserow MIT Expat · MinIO AGPL-3.0 — AGPL is already precedent via ownCloud | known | W40 D3 |
7 — Scoped negatives — what this doc does NOT claim
| ❌ Not claimed | Why |
|---|---|
| "The POC is deployed / running" | No instance deployed in any record; all results are document/source/registry reads |
| "Baserow guarantees no training on our data" | Their §7.2 covers their provider APIs; self-host + BYOK ⇒ the contract is ours (§4.4-3) |
| "The free tier is enough" | Enough for data + agent access; not enough for governance (RBAC/SSO/audit = Advanced) |
| "MCP is safe by default" | MCP includes Delete rows; Baserow's own guidance requires human verification of destructive ops |
| "ATL1 == MEM1" | Different data-plane feature sets; region choice changes the architecture |
| "Email works" | Proton is the interim, Twilio is the target; SMTP wiring + token mint are open items (§9) |
| "Sign-up is disabled" | A directive (a3b4179b), not a verified state |
Any running-version claim for baserow.weown.tools |
Not probed |
| Any byte count on a rolling page | Pricing/docs pages move between reads; cited by sha + clock |
| Any CCC-ID minted by this seat | CCC-ID = 🐳 lane (R-194); referenced only |
8 — POC charter — test plan, success criteria, exit decision
8.1 POC question
Can the Always-Free self-hosted Baserow OSE serve as #SharedBrain pillar ③ — a database a non-technical co-op member can operate AND an agent can responsibly address — running on our own v5 DOKS, inside our DNS/TLS/email posture, at a governance cost we accept?
8.2 Test plan
| # | Test | Method | Answers |
|---|---|---|---|
| T1 | Deploy | helm install pinned chart 1.1.1 on v5 DOKS (region per §6-10 decision) with baserow.weown.tools 3-hostname DNS-only recipe; decide BYO vs bundled (bitnamilegacy trap named) |
W1 · Test A |
| T2 | Peer review | Review @SHD's v5 DOKS deployment as the host platform (origin W2) | W2 |
| T3 | Free-tier value | Exercise Always Free against real co-op workloads for 2 weeks: databases, forms, dashboards, automations, portals | Test B · GTM's live question |
| T4 | Agent seam | Connect one MCP client per endpoint (per §4.2 credential discipline); run list/create/update against a scratch database; verify destructive-op gating manually | §4.1 · §4.4-1 |
| T5 | Governance probe | Attempt the collaboration patterns we actually need (read-only member, attributable edit history, invite flow) and record exactly which fail behind Advanced | §4.4-2 (the tension) |
| T6 | Email posture | Invite delivery through Proton interim; prepare Twilio target (DKIM already live) | §6-7 |
8.3 Success criteria
| # | Criterion | Measure |
|---|---|---|
| S1 | Runs on our DOKS | Pinned chart healthy at baserow.weown.tools, TLS via Caddy ACME, DNS-only |
| S2 | Non-techie operates it | @PLT (SME) + one co-op-level user complete a real task without SQL or a developer |
| S3 | Agent addresses it safely | MCP client CRUD against a scratch DB, with per-endpoint credentials + one documented revocation |
| S4 | Governance cost measured | T5 findings written up: every failed collaboration pattern named, with the licence row that gates it |
| S5 | Email delivers | Invite lands in inbox via Proton interim (target Twilio prepared) |
8.4 Exit decision (options — @GTM R-011 owns the pick)
| Option | When it wins |
|---|---|
| E1 — ADOPT on Always Free | T5 shows the governance gap is tolerable for POC-phase collaboration (invite-only + gatekeeper accounts cover it) |
| E2 — ADOPT + buy Advanced | RBAC/SSO/audit prove load-bearing and the per-user price is accepted against the anti-extractive posture |
| E3 — REPLACE | T3/T5 fail and an audited alternative (Grist/Appsmith/Teable/plane-CE) passes the same tests — note none is runtime-tested yet (§3.2) |
| E4 — DROP pillar ③ | The non-techie UX gap is better served by staying on Supabase + ownCloud alone |
9 — Open items
| # | Item | Owner | State |
|---|---|---|---|
| 1 | R-011 on this POC doc (review → finalize → accept) | @GTM | ⬜ open |
| 2 | Deploy decision: region (ATL1 vs MEM1) + bundled vs BYO object store | @GTM · @PLT · @SHD | ⬜ open |
| 3 | Sign-up-off implementation (directive a3b4179b → verified state) |
@PLT | ⬜ open |
| 4 | SMTP token mint — Proton interim sender (noreply@/baserow@weown.net) + Twilio account alignment (u111720651 vs u112197143) |
@GTM · @PLT · @MWK | ⬜ open |
| 5 | Re-baseline the P0 window (stated W41; today W42 D4) | @GTM · @PLT | ⬜ open |
| 6 | SOT path + target repo for this doc (predecessors disagree: WeOwnBuzz/s004 :: _CHANNELS_/ vs WeOwnSeason005/pocs) |
@GTM | ⬜ ruling |
10 — Provenance & review status
| Field | Value |
|---|---|
| Drafter | 🌾 CCC-MAIT-S004-GTM-MiMo (Xiaomi MiMo-V2.6-Pro) — lane [1.1] |
| Backup pen | 🛡️ CCC-MAIT-S004-GTM-GLM — lane [1.2], activates only on declared unresolved errors |
| Review panel | 🐝 + MC-BA-CCC-META + CCC-GTM-CORE6 (🐳 💻 🛡️ 🟩 🚀) — bar = 🐝 contract G1–G10 |
| CCC-ID | 🐳 lane (R-194) — this seat never mints |
| Review state | DRAFT v0.1 posted for REVIEW — gate ② open |
| R-011 | ⛔ NOT charged (#127 — stage only) |
✅ BP-075 CANONICAL HASH — 🌾 DRAFT v0.1 [drafter-minted · bp075_canonical_v2 · stage-only, never push | #HITL by @yonks at 04:28 EDT on 15Oct2026]
Content-SHA256: 97f8a5a507c522fad0f7dc57c50d3875de407c19984b65a8625644d4771dd72d FEDARCH-CANARY: 97f8a5a5 CHARACTERS: 23723 WORDS: 3711 LINES: 313
♾️ WeOwnNet 🌐 [DUNA][.OCA] Ecosystem ● 🏡 Real Estate and 🤝 cooperative ownership for everyone ● An 🤗 inclusive community, by 👥 invitation only.