From 54f27b50431ff1f731e0ec35b687e00966e7b2a9 Mon Sep 17 00:00:00 2001 From: Rafael Pires Date: Tue, 8 Sep 2026 20:40:56 -0300 Subject: [PATCH] docs: record Stage 3 opening (2026-05) in CLAUDE.md and DECISIONS.md CLAUDE.md still listed Stage 3 as 'Next' and 'external integrations' as removed, four months after v1.0.6 opened Stage 3 with the Datadog DORA integration (#15, docs/PLAN-datadog.md). Update the stage table, narrow the non-goal to billing/webhooks with the Datadog precedent named, adjust the transition line, and add a retroactive ADR (dated 2026-05-13, recorded 2026-09-08) formalizing what CHANGELOG and PLAN already state. Co-Authored-By: Claude Opus 5 (1M context) --- CLAUDE.md | 8 ++++---- docs/DECISIONS.md | 25 +++++++++++++++++++++++++ 2 files changed, 29 insertions(+), 4 deletions(-) diff --git a/CLAUDE.md b/CLAUDE.md index a6988d4..4c03d6a 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -30,10 +30,10 @@ All development should still trace back to the core question: **does this help u |---|---|---| | 0 | Signal Discovery | Complete — 30 analysis modules validated on 58 repos, 3.5k commits, 1.2k PRs | | 1 | Attribution & Adoption | Substantially complete — `prepare-commit-msg` hooks, `iris hook install`, co-author policy guide, CLI push with OAuth | -| 2 | Intelligence Platform (MVP) | **Current** — Next.js + Supabase + NextAuth multi-tenant app ingesting real metrics via `POST /api/ingest`, dashboard + repos + compare views shipping | -| 3 | Scale & Enterprise Readiness | Next — SSO, fine-grained RBAC, cross-system correlation (CI, incidents), anonymized benchmarking | +| 2 | Intelligence Platform (MVP) | **Current** — Next.js + Supabase + NextAuth multi-tenant app ingesting real metrics via `POST /api/ingest`, dashboard + repos + compare views shipping; refinement ongoing | +| 3 | Scale & Enterprise Readiness | **Opened 2026-05** with cross-system correlation (Datadog DORA events — CHANGELOG v1.0.6, `docs/PLAN-datadog.md`, ADR 2026-05-13 in `docs/DECISIONS.md`). Still pending: SSO, fine-grained RBAC, anonymized benchmarking | -Work that lands should fit Stage 2 refinement or prepare Stage 3 — not re-open Stage 0 assumptions without justification. +Work that lands should fit Stage 2 refinement or Stage 3's opened integration capability — not re-open Stage 0 assumptions without justification. --- @@ -67,7 +67,7 @@ Do NOT introduce: - dashboard sprawl: each new view must tie to a validated insight, not a hypothetical user - microservices, event buses, or distributed queues in the platform - enterprise abstractions (RBAC matrix, policy engines, SCIM) before Stage 3 is opened -- billing, webhooks, or external integrations (the original SaaS scaffold had these; they were intentionally removed) +- billing and webhooks (the original SaaS scaffold had these; they were intentionally removed). Analytical data-source integrations (e.g. Datadog DORA events) are Stage 3 scope; the Datadog integration (#15, `docs/PLAN-datadog.md`) is the precedent for how one lands — issue and plan first, then code If unsure, choose the simpler implementation. diff --git a/docs/DECISIONS.md b/docs/DECISIONS.md index cebbd6b..967c6e8 100644 --- a/docs/DECISIONS.md +++ b/docs/DECISIONS.md @@ -274,6 +274,31 @@ Keeping insights on disk artificially caps the audience and blocks the multi-rep --- +## 2026-05-13 — Stage 3 Opens With Integration Capability Only + +*Recorded retroactively on 2026-09-08. The decision itself was taken and shipped with v1.0.6; this entry formalizes what `CHANGELOG.md` and `docs/PLAN-datadog.md` already state, so the repository's context files stop contradicting its shipped state.* + +### Decision + +Stage 3 ("Scale & Enterprise Readiness") is open as of v1.0.6, **scoped to one capability**: consuming analytical data sources for cross-system correlation — first instance, Datadog DORA events (#15). Every other Stage 3 item — SSO, fine-grained RBAC, SCIM, billing activation, webhooks, anonymized benchmarking — remains deferred and needs its own decision. + +### Context + +The 2026-04-12 entry above kept Stage 3 items as "explicit non-goals until Stage 3 is formally opened". `CHANGELOG.md` v1.0.6 (2026-05-13) announced "Stage 3 opens" with the Datadog DORA integration, and `docs/PLAN-datadog.md` described it as "the implicit opening of Stage 3 … only the integration capability". Neither the stage table nor the "external integrations" non-goal in `CLAUDE.md` was amended, so for four months the file every agent session loads instructed against work the project had already shipped. + +### Rationale + +Cross-system correlation (CI, incidents, deploys) is listed under Stage 3 in `CLAUDE.md` and is the natural next input for the durability signals the engine already produces. It does not require the enterprise abstractions (RBAC matrix, policy engine, SCIM) that motivated deferring the rest of Stage 3, so opening the stage narrowly keeps "avoid premature architecture" intact while ending the contradiction. + +### Consequences + +- `CLAUDE.md`: the stage table marks Stage 3 as opened (2026-05) with integration capability; the non-goal is reworded to "billing and webhooks", and analytical data-source integrations are named as Stage 3 scope. +- The Datadog integration (#15 + `docs/PLAN-datadog.md`) is the precedent for how a data-source integration lands: issue and plan first, then code. +- Does **not** open SSO, RBAC, SCIM, billing, webhooks or benchmarking. +- Supersedes the "until Stage 3 is formally opened" clause of the 2026-04-12 entry for integration capability only; leaves that entry's other consequences intact. + +--- + ## 2026-06-11 — Vendor AI Telemetry Allowed Under Privacy-by-Construction (Revises Principle #7) ### Decision