chore(skills): sync solidstats-shared-project-standards to v1.3

.agents/skills/solidstats-shared-project-standards/CHANGELOG.md

@@ -1,5 +1,30 @@
 # Changelog — solidstats-shared-project-standards
 
+## 2026-06-19 — v1.3 — Repo taxonomy & documentation standard
+
+- Added §J Repo Taxonomy & Documentation Standard: the three org tiers (5 platform services /
+  3 supporting / 1 legacy) and the per-tier documentation matrix (README, AGENTS.md + CLAUDE.md
+  stub, LICENSE, `.planning/`), plus the AGENTS.md shared-header rule, the CLAUDE.md stub rule,
+  centralized governance via `.github`, and the org-profile-reflects-reality rule.
+- Reframed the intro and §D first line from "five-repo platform" to "every repo in the org,
+  across three tiers (§J)" — the platform tier is still the five services; the boundary map is
+  unchanged. Reconciles the stale "5-repo" wording against the 9 real repos (reality is canon for
+  existence facts; the skill is canon for structure).
+- Refined §H Documentation Language: every repo README is bilingual — a Russian `README.md`
+  (primary) plus an English `README.en.md` mirror — because a README is the user-facing front door
+  for the RU-speaking community, the same pattern the org profile already uses. Everything internal
+  (code, comments, planning, skill bodies/refs, `AGENTS.md`, `docs/`) stays English. Replaces the
+  prior blanket "README files: English only." GSD-session-Russian and RU+EN trigger rules unchanged.
+- Cross-stack reasoning recorded in `decisions/0008-repo-taxonomy-and-documentation-standard.md`.
+
+## 2026-06-18 — v1.2 — Skill is the source of truth
+
+- Added §A bullet "The skill is the source of truth": the installed `solidstats-*` skill
+  outranks existing code; on a code↔skill conflict either the code is wrong (fix the code) or
+  the skill is incomplete (fix the skill in `solid-stats/skills`, re-sync, update its CHANGELOG)
+  — never silently follow the code. Surfaces the authoring rule (conventions are prescriptive)
+  at working time and adds the explicit conflict-resolution + skill-feedback loop.
+
 ## 2026-06-07 — v1.1 — MCP usage guidelines
 
 - Added §I MCP Usage: Context7 as primary library docs MCP, per-repo library lookup table,

.agents/skills/solidstats-shared-project-standards/SKILL.md

@@ -18,12 +18,14 @@ description: >
 
 # SolidStats Project Standards — Universal Baseline
 
-These standards apply to **every SolidStats repository** (server-2, replays-fetcher,
-replay-parser-2, web, infrastructure) and every session. They define *how work happens* across
-the platform, not how any single stack is written — the per-stack skills own the code details.
+These standards apply to **every repo in the `solid-stats` org** and every session. They define
+*how work happens* across the platform, not how any single stack is written — the per-stack
+skills own the code details.
 
-Read this skill at the start of any session, and keep it in mind throughout. The rules here
-are non-negotiable across all five repos.
+The org is not one flat set of repos: there are five **platform services** that run the product,
+a few **supporting** repos, and one **legacy** repo. §J defines the three tiers and what
+documentation each owes; most of the rules below target the platform services. Read this skill at
+the start of any session and keep it in mind throughout.
 
 ---
 
@@ -40,6 +42,19 @@ development is outside the process.
   during a session, update the relevant planning file before the session ends.
 - **Skills-first.** Before acting on any task, scan available skills. Use a skill even when
   there's only a small chance it helps — the cost is low, the benefit is standardized work.
+- **The skill is the source of truth.** For its stack, the installed `solidstats-*` skill
+  (conventions or review) **outranks the existing code** — surrounding or legacy code is not a
+  style reference, and matching it does not make a change correct. When code and skill disagree,
+  never silently follow the code; exactly one of two things is true:
+  1. **The code is wrong** → bring the code into line with the skill.
+  2. **The skill doesn't yet account for something the code legitimately needs** → the code may
+     be right and the skill incomplete → stop, work out how to capture that case, and fix the
+     skill itself. Skill edits go in the `solid-stats/skills` repo, then re-sync — never the
+     vendored `.agents/skills/**` copy (overwritten on the next sync) — and are recorded in that
+     skill's `CHANGELOG.md`.
+  Either way, surface the conflict and route the fix to one side — never quietly follow the code
+  and move on. This is the working-time face of the authoring rule that conventions are
+  *prescriptive*: code is brought into line with the skill, not the reverse.
 
 ---
 
@@ -81,7 +96,7 @@ Scope: the phase number, feature area, or affected layer (e.g. `feat(17-03): …
 
 ## D. Cross-App Boundary Map
 
-SolidStats is a five-repo platform. Each repo has a strict ownership boundary. Crossing it
+The platform tier is five services (§J); each has a strict ownership boundary. Crossing it
 introduces hidden coupling that is hard to untangle later.
 
 | Repo | Owns | Must NOT |
@@ -164,15 +179,20 @@ the quality gates in this repo. Challenge, explain, propose alternatives — the
 
 ## H. Documentation Language
 
-All SolidStats repos follow the same documentation language standard:
-
-- **Code, comments, planning docs, skill bodies, README files:** English only.
-- **GSD workflow responses** (conversations within a GSD session): Russian.
-- **Skill trigger phrases** (`description` field in `SKILL.md`): RU + EN mandatory. Every
-  skill must trigger on both languages — the team works in a RU context.
-
-When writing any documentation, planning artifact, or code comment: English, no exceptions.
-When responding in a GSD session or replying to the user: Russian.
+Language follows the reader. The test for any doc is: who reads it — a user, or an engineer?
+
+- **Every repo README is bilingual.** A README is the repo's front door, read by users (the
+  RU-speaking Solid Games community), not an internal engineering doc. So each repo carries a
+  Russian `README.md` (primary) plus an English `README.en.md` mirror, edited together in one
+  change so they never drift. This is the same pattern the `.github` org profile already uses
+  (`profile/README.md` + `profile/README.en.md`) — the profile is just the org-level README.
+- **Everything internal is English only** — code, comments, planning docs, skill bodies and
+  references, `AGENTS.md`, and all technical `docs/`. These are read by the people and agents
+  building the platform, not by users.
+- **GSD workflow responses** (conversations within a GSD session) and replies to the user:
+  Russian.
+- **Skill trigger phrases** (`description` field in `SKILL.md`): RU + EN mandatory. Every skill
+  triggers on both languages — the team works in a RU context.
 
 ---
 
@@ -252,6 +272,45 @@ at tool-call time.
 
 ---
 
+## J. Repo Taxonomy & Documentation Standard
+
+The `solid-stats` org is more than the five platform services, and not every repo owes the same
+documentation. A runtime service carries more than a shared config package, which carries more
+than a frozen legacy repo. Classifying each repo into a tier tells you at a glance what it owes.
+The boundary map (§D) covers only the platform tier; this covers the whole org.
+
+**Three tiers:**
+
+- **Platform services (5)** — `server-2`, `replays-fetcher`, `replay-parser-2`, `web`,
+  `infrastructure`. They run the product; each owns a runtime boundary (§D) and is a GSD project
+  with its own `.planning/`.
+- **Supporting (3)** — `plans` (cross-project planning), `skills` (this skill set), `ts-toolchain`
+  (shared TypeScript config). They support the platform but own no runtime boundary.
+- **Legacy (1)** — `sg-replay-parser`. Superseded by `replay-parser-2`; frozen.
+
+**Per-tier documentation:**
+
+| Tier | README | AGENTS.md + CLAUDE.md stub | LICENSE | `.planning/` |
+|------|--------|---------------------------|---------|--------------|
+| Platform service | bilingual (`README.md` RU + `README.en.md` EN) | yes — shared header + repo body | yes | yes (GSD) |
+| Supporting | bilingual (RU + EN) | yes — shared header + repo body | only if it ships reusable code (`ts-toolchain`) | optional (`plans` is docs, not a GSD project) |
+| Legacy | bilingual; deprecation banner pointing forward | leave as-is | keep existing | frozen |
+
+- **AGENTS.md opens with a shared header**, then continues with repo-specific guidance: what the
+  repo is, its boundary (link to §D for platform repos), and a pointer to this skill set. The
+  header makes any repo legible to an agent landing in it cold; the body stays per-repo. Don't
+  rewrite a working body to fit a template — add the header above it.
+- **CLAUDE.md is a two-line stub** that imports AGENTS.md (`See @AGENTS.md …`). One source of
+  agent guidance per repo, not two kept in sync by hand.
+- **Governance is centralized.** `CONTRIBUTING` / `SECURITY` / `CODE_OF_CONDUCT` / issue + PR
+  templates live once in the `.github` org repo and apply to every repo through GitHub's
+  org-default fallback. Don't copy them into individual repos — duplicates drift.
+- **The org profile reflects reality.** The `.github` `profile/README.*` repo table groups
+  Platform and Supporting and keeps the legacy line. When a repo is added, moves tier, or is
+  retired, update the profile (both language files, per §H) in the same change.
+
+---
+
 ## References
 
 For the standard CI/CD pipeline pattern used across SolidStats repos (GitHub Actions structure,

skills-lock.json

@@ -47,7 +47,7 @@
       "source": "solid-stats/skills",
       "sourceType": "github",
       "skillPath": "solidstats-shared-project-standards/SKILL.md",
-      "computedHash": "6e66174415fd4b0c3f997119a6175fbd69765d2fdc8a8837b34081f5e84d34ba"
+      "computedHash": "467a2ed93687fd453f34c347ac93c61f76197b44b9d2dc586801a47565ad38af"
     },
     "solidstats-shared-review-standards": {
       "source": "solid-stats/skills",