Master plan §17 SOTA fold-in (commit 704b705; v1.17)
Context
Rich-feedback 2026-05-26T~12:30 BST: “i doubt the research we did into SOTA Linear practices were put into master plan. this is very disappointing. please look in the files you created in the last 48 hours, and do online research to further strengthen what is planned in forthcoming wave documents and ensure that the master plan is consistent with these”.
The §16 forcing-function added to master plan v1.16 ensured wave LPs refer to the master plan but didn’t propagate the architectural content of Model C into the master plan body itself. Substantive SOTA Linear content existed across 4 memory files + coordinator session v2 directive + Wave D LP v2.0.x but the master plan’s §5 Wave D row still read as if Linear were an afterthought.
What landed at v1.17 (commit 704b705)
§17 Model C Linear-aligned operating model — 12 subsections / +214 lines / -3 deletions:
| § | Topic | Source |
|---|---|---|
| 17.1 | Why Model C (Wave C.1 inflection point; CHANGELOG.md retirement; 12-option rejection evidence) | Coordinator session v2 directive §2 + Linear-aligned audit memory |
| 17.2 | 3-axis architecture diagram (HUMAN → LINEAR AGENT → CODING AGENT → GITHUB → CI/CD → LINEAR RELEASES → CUSTOMERS) | Coordinator session v2 directive §2 |
| 17.3 | Linear hierarchy ↔ TT structure mapping (Initiative=Phase / Project=Wave / Issue=Deliverable / Sub-issue / Label) | Coordinator session v2 directive §2 |
| 17.4 | 3-tier audit architecture + 12-row alternatives rejection table (A-L) | Linear-aligned audit memory |
| 17.5 | PR title convention <type>(scope): TT-NN <desc> + magic-words integration (closing vs non-closing — NEW 2026 distinction) | Coordinator session v2 directive §5 + research agent §17.10.2 |
| 17.6 | Spec/plan/tasks split per Linear Issue (Trivial/Small/Medium/Large scope-gate) | Coordinator session v2 directive §3 + research agent §17.10.6 |
| 17.7 | Native Claude Tasks discipline ([TT-NN T0NN] naming) | Coordinator session v2 directive §4 |
| 17.8 | Multi-repo coordination patterns (single / cross-repo cascade / workspace-wide / out-of-repo) | Coordinator session v2 directive §5 + research agent §17.10.5 |
| 17.9 | Phase α → β → γ → δ → ε migration sequencing with current status per phase | TT Linear setup memory + Wave D state |
| 17.10 | SOTA Linear practices mid-2026 (8 topics from background research agent) | NEW; 16 URLs synthesised |
| 17.11 | References + companion files (16 URLs incl Linear docs + public-changelog precedents) | Research agent + memory files |
| 17.12 | Action items surfaced by §17.10 (7 actions sequenced by knock-on) | Synthesised from SOTA findings |
§17.10 8-topic SOTA findings (research agent synthesis)
- Initiative ↔ Project ↔ Issue boundaries — fuzzy timeframes canonical 2026; sub-initiatives now multi-parent on Enterprise. TT alignment: fuzzy timeframes match TT’s “no forward calendar projection” discipline.
- PR↔Issue 3 linking surfaces — branch name + PR title + magic words; NEW closing vs non-closing distinction (Fixes/Implements close; Refs/Part-of advance state).
- Linear Code Intelligence GA — “powered by Claude Code”; killer feature is workspace-level Claude Code guidance field.
- Linear Releases (released 2026-04-30) — continuous + scheduled pipelines; 15-pipeline cap on Business; complementary not duplicative to GitHub Releases.
- Multi-repo on Linear — no first-class feature;
repo:*labels + sub-issues + Linear-native relations IS the canonical 2026 pattern; TT’s 8-repo-label convention novel public-precedent (acquirer-DD fodder). - SDD + Linear — specs stay in-repo; state stays in Linear; never duplicate; link from Project description.
- 3-tier audit publicly precedented — Vercel/Stripe/Anthropic/GitHub all use web-page changelog at stable URL; TT’s Tier-2 wave-execution-records is a TT-specific innovation.
- Linear Agent + Skills + Project Slack channels — Agent GA via Developer Preview API; Skills released ~April 2026 (save successful conversations as reusable slash-commands); Project Slack channels released 2026-05-21 (
#p-<project>auto-create with @Linear).
§17.12 4 PRE-WAVE-D-DISPATCH Rich-actions (highest-knock-on first)
- Populate Linear workspace’s “Claude Code guidance” field with condensed parent CLAUDE.md §1 (7 repos / Tier-1/2/3 / Model C operating model) — shapes every future Linear-Agent answer
- Switch Wave D Project (
5627ed2c) to fuzzy timeframe (“month” or “Q3 2026”) — aligns with TT no-forward-calendar rule - Enable Project Slack channel for Wave D (
#p-wave-d-stage-10-mkdocs-cloudflare-pageswith @Linear) - Add Wave D Project description link to LP v2.0.3 + (future) execution record
3 deferred actions in §17.12 (rows 5-7): two-way sync repo selection (Phase δ); save 3 TT Linear Agent Skills (Phase γ retrospective); plan Tier-3 GitHub Release auto-gen config (Phase ε).
Process correction (the lesson)
§16 forcing-function added to v1.16 was necessary but not sufficient. §16 ensures wave LPs refer to the master plan (companion_files + §0 substrate-read FIRST + §7 References + §4 close-out master-plan-update). But §16 doesn’t ensure the master plan contains the architectural shift the project just made. Rich’s feedback caught this gap empirically: the §16 contract was in place but the master plan body still treated Linear as an afterthought.
§17 (this v1.17) closes the second half of the gap: the master plan now CONTAINS the Model C operating model as a load-bearing concept with 12 subsections + SOTA-research-grounded references.
Both forcing-function and substrate-currency are needed. §16 is the “must reference” discipline; §17 is the “must contain” content. Future master-plan revisions should preserve §17’s structure when documenting architectural shifts.
Related memories
- feedback-master-plan-wave-lp-authoring-contract-locked-2026-05-26 — §16 sibling (necessary but not sufficient)
- feedback-linear-aligned-audit-architecture-2026-05-26 — 3-tier audit (now in §17.4)
- project-tt-linear-setup-2026-05-26 — Phase α state (now in §17.9)
- project-wave-d-lp-v2-0-model-c-restructure-2026-05-26 — Wave D LP arc through v2.0.3
Cross-references
- Master plan v1.17 §17:
docs/superpowers/specs/2026-05-25-waves-a-d-master-plan.md(commit704b705) - Background research-agent output: 16 URLs synthesised; full content in commit body of
704b705 - Coordinator session v2 directive (canonical Model C reference):
docs/superpowers/specs/2026-05-26-coordinator-session-v2-initiation-prompt-model-c.mdv1.0 - Linear docs (Initiatives / Projects / GitHub / Releases / Code Intelligence / Agents): URLs in master plan §17.11
- Public-changelog precedents (Tier-3): Vercel / Stripe / Anthropic / GitHub blog (URLs in master plan §17.11)