Snapshot WIP: solver HP epic progress, BPHX/HX physics, BMAD skill refresh.

Capture uncommitted solver robustness work (regularization, domain errors, linear solver lifecycle, tube DP/MSH), web workbench updates, and synced BMAD skills across IDE agent folders before starting BPHX pressure-drop.

Co-authored-by: Cursor <cursoragent@cursor.com>

.agent/skills/bmad-architecture/references/headless.md

@@ -0,0 +1,26 @@
+# Headless
+
+No interactive user: infer everything, ask nothing, but never invent — record inferences as `assumptions[]` and gaps that need a human as `open_questions[]`. Detect headless from a `headless: true` flag, a non-interactive / no-TTY invocation, an activation hook that declares it, or a first message that pre-supplies all inputs and asks for an artifact path back; when ambiguous, default to interactive.
+
+Drive the run from the payload in the first message — `intent`, `altitude`, `purpose`, the driving input (spec package / PRD / raw intent / brownfield path), a parent spine path at lower altitude, and `doc_workspace` if a specific folder is required. Infer anything absent from the inputs or workspace; don't invent stack, constraints, or scope to fill a gap. You still verify named tech on the web (you can't ask, but you can check) and still drive every write through the shared `{project-root}/_bmad/scripts/memlog.py`. Run the full Reviewer Gate (`references/reviewer-gate.md`) non-interactively: `scripts/lint_spine.py` plus **every `{workflow.finalize_reviewers}` lens as a parallel subagent** (and any ad-hoc lens the spine's criticality warrants). Headless skips only the human picking from the menu — never the reviewers themselves; apply the clear fixes and record anything unresolved in `open_questions[]`. For a true authority collision, list it in `conflicts_with_prior_decisions[]`. For the Validate intent, always write the report to `{doc_workspace}` and add `"offer_to_update": true`. If intent stays ambiguous after inference, halt blocked.
+
+End with JSON only, omitting keys for artifacts not produced — the shape below is the fully-produced (`complete`) case; a `blocked` run produces no spine, so it omits `spine`, `memlog`, and `companions` entirely (see the note under the block):
+
+```json
+{
+  "status": "complete | partial | blocked",
+  "intent": "create | update | validate",
+  "altitude": "initiative | feature | epic",
+  "purpose": "build-substrate | discussion",
+  "doc_workspace": "<resolved run folder>",
+  "spine": "{doc_workspace}/ARCHITECTURE-SPINE.md",
+  "memlog": "{doc_workspace}/.memlog.md",
+  "companions": [],
+  "assumptions": [],
+  "open_questions": [],
+  "conflicts_with_prior_decisions": [],
+  "reason": "<one line, only when blocked>"
+}
+```
+
+`complete` stands alone · `partial` (spine produced, but `open_questions[]` non-empty or critical inputs inferred) means review before downstream use · `blocked` means no spine produced — return only `status`, `intent`, `reason`, and `doc_workspace` (if bound), omitting `spine`, `memlog`, `companions`, and the artifact arrays that don't exist.

.agent/skills/bmad-architecture/references/reviewer-gate.md

@@ -0,0 +1,13 @@
+# Reviewer Gate
+
+The spine's pre-handoff review. Runs at Finalize (after distill + reconcile) and *is* the Validate intent. The difference is the ending: at Finalize you apply the clear fixes yourself; under Validate you report and don't change the spine.
+
+Cheap deterministic pass first: `uv run {skill-root}/scripts/lint_spine.py --workspace {doc_workspace}` settles the mechanical misses (placeholders, duplicate `AD` IDs, missing Binds/Prevents/Rule, unpinned Stack versions), so reviewers spend judgment on the semantic half.
+
+Assemble the menu: a **rubric walker** that judges the spine against the good-spine checklist below, **+ every entry in `{workflow.finalize_reviewers}`**, + ad-hoc lenses you invent or offer as the spine's rigor, altitude, and criticality warrant — a security/compliance lens for regulated stakes, a seam reviewer cross-team, a data-integrity lens for a heavy data model. Scale *whether and how heavily the gate runs* to the stakes: a throwaway prototype may run it quietly or skip the gate entirely; a high-criticality or platform-altitude spine earns more lenses and the explicit all / subset / skip menu. But once the gate runs, the `{workflow.finalize_reviewers}` always run — they are the configured floor, never cherry-picked out; only the ad-hoc lenses are optional. (Headless never skips the gate.)
+
+Dispatch every entry as a **parallel subagent against `ARCHITECTURE-SPINE.md`** (prefix convention: `skill:` / `file:` / plain text). Each writes its full review to `{doc_workspace}/reviews/review-{slug}.md` — a subfolder, so the gate's scratch stays out of the deliverable folder — and returns ONLY a compact summary (verdict, top 2–5 findings, file path) — the parent never holds full review text. An inline self-check does not count: the independent context is the point, because a fresh reviewer finds the divergences the author talks past. If subagents are unavailable, run sequentially — write the file first, then flush it from context.
+
+**Good-spine checklist** (what the rubric walker judges): it fixes the real divergence points for the level below and misses none; every `AD`'s Rule is enforceable and actually prevents its stated divergence; nothing under Deferred could let two units diverge; named tech is verified-current; it ratifies rather than contradicts a brownfield codebase; if a spec drove it, it covers that spec's capabilities; if a parent spine is inherited, no new `AD` weakens or contradicts an inherited one; and every dimension the altitude owns is decided, deferred, or an open question — a whole dimension left silent is a finding, especially the operational/environmental envelope (deployment & environments, infra/provider strategy, operations) a domain-focused draft skips.
+
+Surface findings tiered, never dumped: a one-sentence gate verdict, then critical + high; medium/low roll into a tail ("plus N more in {file}"). Per finding: autofix, discuss, defer to Deferred / open items, or ignore. **At Finalize this is your own gate — apply the clear fixes rather than handing over a list; surface only what genuinely needs the user.** Under the **Validate intent**, fold every reviewer's output into one bespoke HTML + markdown report and open the HTML.