@event4u/agent-config 16.5.0 → 17.0.0

package/dist/agent-src/skills/roadmap-writing/SKILL.md

@@ -155,7 +155,7 @@ gate fires later).
 ### 4c. Autonomy-first — zero human gates by default
 
 Canonical rule:
-[`templates/roadmaps.md` rule 22](../../agent-src/templates/roadmaps.md).
+[`templates/roadmaps.md` rule 22](../../templates/roadmaps.md).
 Default human-checkpoint count: **zero**. Every step is
 agent-executable — `- [ ] User verifies X` steps and "Review /
 Sign-off" phases are authoring bugs; replace each with an
@@ -190,8 +190,8 @@ seven fields of rule 20), never a stray "blocked on X" sentence. Write it so
 the owner can decide in one sitting: one option named and why, what the delay
 costs, a command or path per option rather than prose, and an offer to walk
 them through it. Ratcheted by `lint_roadmap_blockers`. Full shape:
-[`templates/roadmaps.md` rule 20](../../agent-src/templates/roadmaps.md).
-Omit it entirely when there is no such gate; run the § 4c gate-test first. A **hold** is the publication-side twin — a blocker stops execution, a hold stops publication; ask once whether an intermediate tree state must stay unpublished (default **no**; roadmap incompleteness never is one), walk `re-sequence → guard → hold`, and record the avoided outcome as well as the taken one, per [`templates/roadmaps.md` rule 28](../../agent-src/templates/roadmaps.md) and the § 5c self-check in [`/roadmap:create`](../../domains/product-basic/roadmap/create/command.md).
+[`templates/roadmaps.md` rule 20](../../templates/roadmaps.md).
+Omit it entirely when there is no such gate; run the § 4c gate-test first. A **hold** is the publication-side twin — a blocker stops execution, a hold stops publication; ask once whether an intermediate tree state must stay unpublished (default **no**; roadmap incompleteness never is one), walk `re-sequence → guard → hold`, and record the avoided outcome as well as the taken one, per [`templates/roadmaps.md` rule 28](../../templates/roadmaps.md) and the § 5c self-check in [`/roadmap:create`](../../domains/product-basic/roadmap/create/command.md).
 
 ### 5c. Risk review (Gate R1) — after draft, before save
 
@@ -265,7 +265,7 @@ integrate don't dump), resolve contested design in the council *first*
 then author, encode the decision so it survives (Council notes +
 neutral-descriptor Provenance + memory lock), and make "integration, not
 dump" a testable acceptance criterion. Full four-move detail →
-[`roadmap-writing-source-derived`](../../agent-src/contexts/execution/roadmap-writing-source-derived.md).
+[`roadmap-writing-source-derived`](../../contexts/execution/roadmap-writing-source-derived.md).
 
 ## Output format
 

package/dist/agent-src/skills/rtk-output-filtering/SKILL.md

@@ -83,7 +83,7 @@ rtk docker compose ps # compact container status
    completeness-critical commands).
 
 Installation and one-time setup are owned by
-[`/onboard`](../../commands/onboard.md). If the user asks to install rtk
+the setup wizard (`agent-config setup`). If the user asks to install rtk
 outside onboarding, follow the Installation section below and set
 `personal.rtk_installed: true` on success.
 

package/dist/agent-src/skills/rule-writing/SKILL.md

@@ -150,8 +150,7 @@ invoke [`rule-refactor`](../rule-refactor/SKILL.md). Do NOT:
   do its job, the rule is right and the rule set around it is wrong.
 * Raise `FAIL_THRESHOLD` in `scripts/measure_augment_budget.ts` —
   threshold-lift is explicitly forbidden (see the
-  [`validation-budget`](../../rules/validation-budget.md) rule and
-  the `rule-refactor` Iron Law).
+  [`rule-refactor`](../rule-refactor/SKILL.md) Iron Law).
 * Promote an always-rule to auto to dodge the cap if the rule's
   semantics require always-on visibility — that breaks the rule, not
   the budget.

package/dist/agent-src/skills/security/SKILL.md

@@ -7,13 +7,26 @@ workspaces:
   - engineering
 packs:
   - engineering-base
+triggers:
+  - phrase: "password reset"
+  - phrase: "email verification"
+  - phrase: "sign up"
+  - phrase: "social login"
+  - keyword: "login"
+  - keyword: "registration"
+  - phrase: "session id"
+  - phrase: "session cookie"
+  - keyword: "logout"
+  - keyword: "mfa"
+  - keyword: "passkey"
+  - keyword: "oauth"
 ---
 
 # security
 
 ## When to use
 
-Use when implementing authentication, authorization, or any security-sensitive functionality.
+Use when implementing authentication, authorization, or other security-sensitive work.
 
 Do NOT use when:
 
@@ -55,18 +68,41 @@ The procedure below is stack-agnostic. For framework-specific primitives (Larave
 ### Step 3: Review for adversarial
 
 For security-sensitive changes, run [`adversarial-review`](../adversarial-review/SKILL.md).
-Focus on: attack surface, trusting user input, authorization gaps.
+Focus: attack surface, trusted user input, authz gaps.
 
 ## Conventions
 
 → For PHP / Laravel specifics (auth helpers, mass assignment, Blade escaping, CSRF middleware): see guideline `docs/guidelines/php/security.md`.
 → For other stacks, follow the framework's hardening guide and the carve-outs above.
 
+### Account flows — the whole lifecycle, not just the login
+
+Sign-up, verification, reset, credential and email change, MFA, federated
+sign-in and logout are one surface. Before judging any of them, trace the path:
+input → identity lookup → verification → secret handling → state change →
+session → response.
+
+- A public auth response is an **observable contract**, not an error string:
+  status, body category, redirect and cookies must not differ between a known
+  and an unknown identity, and the verifier path runs for both.
+- Compare every secret with the library's constant-time primitive, and respect
+  that primitive's length precondition. Never a general-purpose fast hash for a
+  password. Never an invented token, one-time code or salt scheme.
+- Changing an identity — address, phone, password, factor, passkey, recovery
+  codes, a linked provider — needs fresh re-authentication and a notice on the
+  channel that was already verified.
+
+Ground the specifics before enumerating from memory: the account-lifecycle
+failure classes, their required controls and their negative tests live in
+[`threat-modeling`](../threat-modeling/SKILL.md)'s corpus under the
+`authentication` surface class. Every numeric value in them defers to the
+external authority named in the block below.
+
 ### Crypto, password storage, certificates — route, do not guess
 
 This skill carries no cryptographic parameter, key size, work factor, cipher
 suite, or TLS version floor: such a value is authoritative-looking long after it stops
-being true. Take it from <https://cheatsheetseries.owasp.org/> at the moment you
+being true. Take it from <https://cheatsheetseries.owasp.org/> when you
 need it — never from memory, never from this file. Rationale and reopening
 condition: [ADR-238](../../../docs/decisions/ADR-238-security-content-routes-to-external-authority.md).
 
@@ -75,7 +111,7 @@ condition: [ADR-238](../../../docs/decisions/ADR-238-security-content-routes-to-
 - Verify all user input is validated at the boundary via the framework's primitive — never trust raw request data.
 - Confirm an authorization check exists for every state-changing action.
 - Check that no raw user input reaches SQL, HTML output, shell commands, or template renderers without escaping.
-- Run the project's type-checker — must pass (catches type-safety issues that enable injection).
+- Run the project's type-checker — must pass (type errors can enable injection).
 
 ## Output format
 
@@ -85,7 +121,7 @@ condition: [ADR-238](../../../docs/decisions/ADR-238-security-content-routes-to-
 ## Gotcha
 
 - Validation ensures format, not intent — don't trust input after validation alone.
-- "Throw" vs "boolean" authz APIs behave differently (`Gate::authorize()` throws vs `Gate::allows()` returns bool in Laravel; `CanActivate` in NestJS throws; FastAPI dependencies throw `HTTPException`). Pick based on how the framework expects failure to surface.
+- "Throw" vs "boolean" authz APIs behave differently (`Gate::authorize()` throws vs `Gate::allows()` returns bool in Laravel; `CanActivate` in NestJS throws; FastAPI dependencies throw `HTTPException`). Pick per how the framework surfaces failure.
 - Rate-limit ALL public endpoints, not just login.
 - Never log passwords, tokens, or API keys.
 

package/dist/agent-src/skills/skill-improvement-pipeline/SKILL.md

@@ -243,7 +243,9 @@ finding never silently becomes a hard gate. Both halves, specified in
   Language agents that reinforce from verbal self-reflection on past
   trials. This pipeline adapts Reflexion by gating promotion with a
   review step — learnings only harden into rules/skills after explicit
-  approval, never auto-commit: for a proposal scoped to the package, on
-  the AI council's ratification record (user as last rung, ADR-281); for
-  a change a user asked for, on that user's approval.
+  approval, never auto-commit: for a proposal scoped to the package, a
+  digest-bound AI council record is the required **review** and the owner's
+  authenticated authorization is the **approval**, with a council refusal
+  failing closed (ADR-281 as amended 2026-10-07); for a change a user asked
+  for, on that user's approval.
 

package/dist/agent-src/skills/skill-reviewer/SKILL.md

@@ -93,7 +93,7 @@ The highest-signal content in a skill. Documents failure patterns.
 
 Skill exceeds size limits (see `docs/guidelines/agent-infra/size-and-scope.md`).
 
-**Check:** Review at >400 lines. Strongly consider split at >1200 words / >1500 words.
+**Check:** Review at >400 lines. A word count alone is no split reason — split only material consulted on demand, or a procedure handed out in order (`skill-writing` § Progressive disclosure).
 **Fix:** Extract reference tables, templates, and examples into separate files
 in the skill folder, or split by responsibility.
 

package/dist/agent-src/skills/skill-writing/SKILL.md

@@ -101,7 +101,7 @@ Do NOT create a skill or rule for:
 
 → See `docs/guidelines/agent-infra/size-and-scope.md` for full limits.
 
-* Target: 300–900 words. Review for split above 1200 words. Strongly consider split above 1500 words.
+* Target: 300–900 words. A word or line count is never by itself a reason to split — § Progressive disclosure says when to.
 * Over the 400-line K6 cap → the **router-head contract** in `templates/skill.md`
   applies: entry head (when-to-use · mode table · routing) plus one `tasks/` or
   `references/` file per mode, and the head never inlines more than one mode's
@@ -119,21 +119,36 @@ named one; the split cost a rename, not a decision.
 
 * **Body:** the trigger, the workflow, the ownership statement, and every
   heuristic that changes a write decision. A skill whose body is a map to its
-  own content is loaded and not applied.
+  own content is loaded and not applied. So a split head keeps every such
+  heuristic and all five required sections (`When to use`, `Gotcha`,
+  `Procedure`, `Output format`, `Do NOT`); only step execution moves out.
 * **`references/`:** corpora, lookup tables, long schemas, worked examples —
   read on the path that needs them, never on every activation.
 
-Splitting is not a size escape hatch. The measured estate (n=289, 2026-08-12):
-median 1,077 words, p90 1,867, **only 6 skills above 2,500 and 4 above 3,000**.
-So a skill at 1,500 words is ordinary, and moving prose out of the body to hit
-a number makes it worse. Split when the material is genuinely
-consulted-on-demand; keep the load-bearing part in the body even when that
-leaves the file long.
+Split for one of two reasons only: the material is genuinely consulted on
+demand, or a procedure needs to be handed out in order. Never split to hit a
+number. The measured estate (n=289, 2026-08-12): median 1,077 words, p90 1,867,
+**only 6 skills above 2,500 and 4 above 3,000** — a skill at 1,500 words is
+ordinary, and the load-bearing part stays in the body even when that leaves the
+file long.
 
 A `references/` split does not change the gates that already apply: a
 `token_budget_class: rich` skill still meets its 3,500-token ceiling on the
 body alone.
 
+### Four words: reference, co-requirement, procedure, chain
+
+* A **reference** is context loaded on demand from the skill's own
+  `references/`. It stays one level deep: a reference file links no sibling
+  reference of its own skill, so the head reaches every one.
+  `lint_skill_link_reach` holds today's sibling links as a shrink-only ratchet.
+* **`requires_skills`** is co-availability — the named skill ships wherever this
+  one does. It states no order.
+* A **procedure** is the ordered steps of one skill. Held as data, it is a
+  sidecar beside `SKILL.md` that names only files of the same skill.
+* A **chain between skills** is not supported (the skill-DAG lock, ADR-263).
+  Prose such as "then invoke X" is a hint to the reader, not a chain link.
+
 ## Modes
 
 This skill is a router head. The decision layer above — what goes where, the

package/dist/agent-src/skills/skill-writing/references/section-patterns.md

@@ -1,5 +1,23 @@
 # skill-writing — section patterns
 
+<!-- reference-toc:start (generated by generate_reference_toc — do not edit) -->
+**Contents**
+
+- [Description-optimizer loop (U1 — held-out, not vibes)](#description-optimizer-loop-u1--held-out-not-vibes)
+- [Self-QA loop for output-producing skills (optional pattern)](#self-qa-loop-for-output-producing-skills-optional-pattern)
+- [Known-pitfalls section (optional pattern, tool skills)](#known-pitfalls-section-optional-pattern-tool-skills)
+- [Rationalizations-to-reject section (recommended pattern, security-stop-routed skills)](#rationalizations-to-reject-section-recommended-pattern-security-stop-routed-skills)
+- [Non-negotiable-deliverable section (recommended pattern, adjacent-technology clusters)](#non-negotiable-deliverable-section-recommended-pattern-adjacent-technology-clusters)
+- [Destructive-operation gates belong in the description](#destructive-operation-gates-belong-in-the-description)
+- [Upstream-version-notes section (optional pattern, wrapped fast-moving tools)](#upstream-version-notes-section-optional-pattern-wrapped-fast-moving-tools)
+- [Security-constraints section (required pattern, script-bearing skills)](#security-constraints-section-required-pattern-script-bearing-skills)
+- [Action-reference split (required pattern, safety_mode: strict skills)](#action-reference-split-required-pattern-safety_mode-strict-skills)
+- [Mechanism-teaching section (optional pattern, skills whose subject has one)](#mechanism-teaching-section-optional-pattern-skills-whose-subject-has-one)
+- [Illustrative-not-verbatim marker (optional pattern, reference code)](#illustrative-not-verbatim-marker-optional-pattern-reference-code)
+- [Headline metric + closing report (optional pattern, optimization skills)](#headline-metric--closing-report-optional-pattern-optimization-skills)
+- [Contrastive-example slot (optional pattern, authoring and behavior skills)](#contrastive-example-slot-optional-pattern-authoring-and-behavior-skills)
+<!-- reference-toc:end -->
+
 > Mode body of the [`skill-writing`](../SKILL.md) skill (router-head
 > retrofit, 2026-08-20). Content moved VERBATIM from SKILL.md — load this
 > file when the mode table in SKILL.md routes here.
@@ -277,7 +295,7 @@ moves the headline number by degrading something it does not measure is the
 failure this pattern exists to make visible, and only the author knows where that
 edge is.
 
-## Contrastive-example slot (optional pattern, authoring and behaviour skills)
+## Contrastive-example slot (optional pattern, authoring and behavior skills)
 
 Where a skill's guidance is easy to agree with and hard to apply — tone, phrasing,
 question shape, what counts as evidence — a wrong/right pair does what prose

package/dist/agent-src/skills/source-discovery/SKILL.md

@@ -16,7 +16,7 @@ rule. Hand-off target from [`think-before-action`](../../rules/think-before-acti
 when a task touches **external or expensive** structure. Definitions
 (Evidence Report buckets, provenance, trust tiers, the DB-not-in-codebase
 boundary, the card-worthiness threshold) live in
-[`evidence-discipline`](../../agent-src/contexts/execution/evidence-discipline.md) —
+[`evidence-discipline`](../../contexts/execution/evidence-discipline.md) —
 read it once; this file is the executable procedure.
 
 ## When to use
@@ -165,7 +165,7 @@ bypasses a fresh structural read. A field/endpoint/column/value is still confirm
 against a live source this session. v2 capture is **write-only into gitignored
 intake** (the agent may *suggest* a signal, never silently commit); trust and
 commit are always human-gated. Full model + the three classes + memory tiers:
-[`project-intelligence`](../../agent-src/contexts/execution/project-intelligence.md).
+[`project-intelligence`](../../contexts/execution/project-intelligence.md).
 
 ## Knowledge capture (`api_shape_learned` / `convention_detected` events)
 
@@ -184,7 +184,7 @@ npx tsx node_modules/@event4u/agent-config/src/scripts/emit_knowledge_event.ts -
 
 `/team-knowledge consolidate` turns accumulated events into
 `agents/knowledge/concepts/` pages as a reviewed batch — see
-[`knowledge-pages`](../../agent-src/templates/contexts/knowledge-pages.md).
+[`knowledge-pages`](../../templates/contexts/knowledge-pages.md).
 
 ## Gotchas
 

package/dist/agent-src/skills/sql-writing/SKILL.md

@@ -14,8 +14,8 @@ packs:
 
 > **Grounded corpus:** tuning decisions (indexes, keyset pagination,
 > N+1, trigram search, lock contention) ground via the
-> [`database`](../database/SKILL.md) corpus — `./scripts-run
-> <skills-root>/corpus-grounding/scripts/ground search
+> [`database`](../database/SKILL.md) corpus — `node
+> <skills-root>/corpus-grounding/scripts/ground.mjs search
 > --manifest <skills-root>/database/data/manifest.json "<symptom>"`.
 
 ## When to use

package/dist/agent-src/skills/standards-from-config/SKILL.md

@@ -20,7 +20,7 @@ the config file, and the card only points at it plus distils the 3–5 points an
 agent needs most while writing.
 
 Definitions (Class A/B/C, trust tiers, the v1↔v2 isolation contract) live in
-[`evidence-discipline`](../../agent-src/contexts/execution/evidence-discipline.md);
+[`evidence-discipline`](../../contexts/execution/evidence-discipline.md);
 the parent context mechanism is [`context-document`](../context-document/SKILL.md).
 
 ## When to use
@@ -129,6 +129,6 @@ A Class-A standards card MUST contain, in order:
 
 ## See also
 
-- [`evidence-discipline`](../../agent-src/contexts/execution/evidence-discipline.md) — Class A/B/C, trust tiers, the isolation contract.
+- [`evidence-discipline`](../../contexts/execution/evidence-discipline.md) — Class A/B/C, trust tiers, the isolation contract.
 - [`context-document`](../context-document/SKILL.md) — the parent context mechanism + storage locations.
 - [`source-discovery`](../source-discovery/SKILL.md) — v1 structural discovery (read fresh; Class A never bypasses it).

package/dist/agent-src/skills/storybook-workshop/SKILL.md

@@ -151,6 +151,11 @@ rather than reporting zero violations — an unrun check and a clean check produ
 number and mean opposite things.
 
 **One finding is decidable without a browser**, and it ships beside this skill:
+
+```bash
+node <skills-root>/storybook-workshop/scripts/story_contrast_floor.mjs <story-file> […]
+```
+
 `scripts/story_contrast_floor.ts` computes the WCAG 2.1 contrast ratio between two colours a
 story *declares in its own args* and emits the same `(rule, selector, severity)` shape. It is
 **not axe** and not a substitute for § Validate — it sees no rendered page, so role, focus,

package/dist/agent-src/skills/storybook-workshop/scripts/story_contrast_floor.mjs

@@ -0,0 +1,99 @@
+// src/skills/storybook-workshop/scripts/story_contrast_floor.ts
+import * as fs from "node:fs";
+var AA_NORMAL = 4.5;
+var AA_LARGE = 3;
+var clamp01 = (n) => n < 0 ? 0 : n > 1 ? 1 : n;
+var luminance = (hex) => {
+  const m = /^#?([0-9a-f]{6}|[0-9a-f]{3})$/i.exec(hex.trim());
+  if (m === null) return null;
+  let h = m[1];
+  if (h.length === 3) h = h.split("").map((c) => `${c}${c}`).join("");
+  const channel = (pair) => {
+    const v = clamp01(parseInt(pair, 16) / 255);
+    return v <= 0.03928 ? v / 12.92 : ((v + 0.055) / 1.055) ** 2.4;
+  };
+  const r = channel(h.slice(0, 2));
+  const g = channel(h.slice(2, 4));
+  const b = channel(h.slice(4, 6));
+  return 0.2126 * r + 0.7152 * g + 0.0722 * b;
+};
+var contrastRatio = (fg, bg) => {
+  const a = luminance(fg);
+  const b = luminance(bg);
+  if (a === null || b === null) return null;
+  const [hi, lo] = a >= b ? [a, b] : [b, a];
+  return (hi + 0.05) / (lo + 0.05);
+};
+var declaredPairs = (source) => {
+  const out = [];
+  const exportRe = /export\s+const\s+([A-Za-z0-9_]+)\s*:[^=]*=\s*\{([\s\S]*?)\n\};/g;
+  for (const m of source.matchAll(exportRe)) {
+    const story = m[1];
+    const body = m[2];
+    const fg = /(?:color|fg|foreground)\s*:\s*['"](#[0-9a-fA-F]{3,6})['"]/.exec(body);
+    const bg = /(?:background|bg|backgroundColor)\s*:\s*['"](#[0-9a-fA-F]{3,6})['"]/.exec(body);
+    if (fg?.[1] !== void 0 && bg?.[1] !== void 0) {
+      out.push({ story, fg: fg[1], bg: bg[1] });
+    }
+  }
+  return out;
+};
+var checkStoryFile = (file, floor = AA_NORMAL) => {
+  let source = "";
+  try {
+    source = fs.readFileSync(file, "utf8");
+  } catch {
+    return [];
+  }
+  const out = [];
+  for (const { story, fg, bg } of declaredPairs(source)) {
+    const ratio = contrastRatio(fg, bg);
+    if (ratio === null || ratio >= floor) continue;
+    out.push({
+      rule: "color-contrast",
+      // The story is the addressable unit here — there is no DOM to select into.
+      selector: `story:${story}`,
+      severity: "error",
+      ratio: Math.round(ratio * 100) / 100
+    });
+  }
+  return out;
+};
+var selfTest = () => {
+  let failed = 0;
+  const check = (label, cond) => {
+    if (!cond) {
+      process.stderr.write(`\u274C  ${label}
+`);
+      failed += 1;
+    }
+  };
+  const r = contrastRatio("#000000", "#ffffff");
+  check("black on white is the maximum 21:1", r !== null && Math.abs(r - 21) < 0.01);
+  check("a colour on itself is 1:1", Math.abs((contrastRatio("#777777", "#777777") ?? 0) - 1) < 0.01);
+  check("an unreadable colour is null, not 0", contrastRatio("nope", "#fff") === null);
+  check("shorthand hex is expanded", contrastRatio("#000", "#fff") !== null);
+  process.stdout.write(failed === 0 ? "\u2705  story_contrast_floor: self-test passed\n" : `\u274C  ${String(failed)} failure(s)
+`);
+  return failed === 0 ? 0 : 1;
+};
+var main = () => {
+  const argv = process.argv.slice(2);
+  if (argv.includes("--self-test")) return selfTest();
+  const files = argv.filter((a) => !a.startsWith("--"));
+  const violations = files.flatMap((f) => checkStoryFile(f));
+  process.stdout.write(`${JSON.stringify({ violations }, null, 2)}
+`);
+  return violations.length > 0 ? 1 : 0;
+};
+if (process.argv[1] !== void 0 && process.argv[1].includes("story_contrast_floor")) {
+  process.exit(main());
+}
+export {
+  AA_LARGE,
+  AA_NORMAL,
+  checkStoryFile,
+  contrastRatio,
+  declaredPairs,
+  luminance
+};

package/dist/agent-src/skills/subagent-orchestration/SKILL.md

@@ -116,9 +116,9 @@ Rules:
   continues the parent's task under identical tools and constraints (it then
   reads the parent's cache on its first request); a named subagent is
   preferred the moment isolation, a different tool set, or nested dispatch is
-  needed — a fork cannot nest, and forking forces background mode, which
-  changes the tool set and invalidates the very prefix that motivated the
-  fork. `do-in-parallel`-shaped fan-out from one parent has this shape by
+  needed — a fork cannot nest. Where the host row says `fork_shares_cache:
+  true` a background fork keeps the parent's prefix and tools; elsewhere it
+  is unseen. `do-in-parallel`-shaped fan-out from one parent has this shape by
   construction (a known, unfixable upstream cost — see
   [`prompts/README.md`](prompts/README.md) § Prompt-cache discipline). NEVER
   encode "always fork" — that was proposed and cut (council 2026-07-30):
@@ -134,7 +134,7 @@ composition and activation **within** the form the static gate already
 picked — severity never overrides the form gate, the Iron Law, or any
 safety floor. Guidance, not a new object class (persona-catalog
 disposition). Severity→composition table + escalation rule →
-[`subagent-modes-detail` § Severity-conditioned team composition](../../agent-src/contexts/execution/subagent-modes-detail.md).
+[`subagent-modes-detail` § Severity-conditioned team composition](../../contexts/execution/subagent-modes-detail.md).
 
 ## The nine modes
 
@@ -146,12 +146,12 @@ model pairing. Defaults come from
 
 Descriptive lookup material (per-mode topology table, anti-drift default,
 glossary) lives in
-[`subagent-topologies`](../../agent-src/contexts/execution/subagent-topologies.md) —
+[`subagent-topologies`](../../contexts/execution/subagent-topologies.md) —
 pull it for capacity planning; it is metadata, not runtime-enforced.
 
 Per-mode decision rows (when to use / when not / model pairing) and the
 mode-2 stage-routing contract live in
-[`subagent-modes-detail` § Modes 1–6](../../agent-src/contexts/execution/subagent-modes-detail.md) —
+[`subagent-modes-detail` § Modes 1–6](../../contexts/execution/subagent-modes-detail.md) —
 pull them at dispatch time.
 
 ### 1. do-and-judge
@@ -202,7 +202,7 @@ leaks into the next. Use for a multi-step cross-wing chain (≥2 senior
 skills, each ≥30 min); not for fast iteration under 30 min (overhead
 dominates). Full handoff shape, example chain, competitive per-candidate
 isolation, and the no-auto-merge Hard Floor →
-[`subagent-modes-detail`](../../agent-src/contexts/execution/subagent-modes-detail.md) § Mode 7.
+[`subagent-modes-detail`](../../contexts/execution/subagent-modes-detail.md) § Mode 7.
 
 ### 8. do-with-live-app-judge (gated — UI-heavy tasks)
 
@@ -212,7 +212,7 @@ reading the diff. Use for UI-heavy change where "looks right in the diff"
 ≠ "works in the app"; not for backend/logic (a diff judge is cheaper).
 Experimental until `verdict_changed_outcome` telemetry proves it. Rubric,
 adoption gate, and the async-verifier future candidate →
-[`subagent-modes-detail`](../../agent-src/contexts/execution/subagent-modes-detail.md) § Mode 8.
+[`subagent-modes-detail`](../../contexts/execution/subagent-modes-detail.md) § Mode 8.
 
 ### 9. adversarial-verification-council (gated — opt-in, advisory)
 
@@ -226,7 +226,7 @@ Mode 6 it emits a **findings-union**, not a go/no-go verdict. **Advisory only
 never auto-gates (Hard Floor).** Default-off (`subagents.adversarial_council`);
 opt-in high-risk changes only; registered claim + high-risk tier need
 cross-*vendor* skeptics. Invariants, skeptic prompt, reconciliation, prove-or-drop
-gate → [`subagent-modes-detail`](../../agent-src/contexts/execution/subagent-modes-detail.md) § Mode 9
+gate → [`subagent-modes-detail`](../../contexts/execution/subagent-modes-detail.md) § Mode 9
 + [`prompts/adversarial-verification-council.md`](prompts/adversarial-verification-council.md)
 
 > **Disposition (2026-07-28, honest null).** The adversarial-council finding-
@@ -246,7 +246,7 @@ four statuses — `DONE` · `DONE_WITH_CONCERNS` · `NEEDS_CONTEXT` ·
 `BLOCKED` — no free-form alternatives; orchestrators route on status
 mechanically. Meaning/required-keys table, why-fixed rationale, and the
 `NEEDS_CONTEXT`-vs-`BLOCKED` distinction →
-[`subagent-modes-detail` § Status taxonomy](../../agent-src/contexts/execution/subagent-modes-detail.md).
+[`subagent-modes-detail` § Status taxonomy](../../contexts/execution/subagent-modes-detail.md).
 
 **The envelope is the ONLY return channel** (token-economy-dispatch
 Phase 6). A worker writes its full output to disk (runtime artifact dir,
@@ -267,7 +267,7 @@ Edits to a prompt do not bloat this skill against the 400-line sunset
 trigger. Eight prompt files cover modes 1–7 and 9 (the standalone judge
 reuses [`prompts/do-and-judge.md`](prompts/do-and-judge.md); mode 8's
 live-app rubric lives in
-[`subagent-modes-detail` § Mode 8](../../agent-src/contexts/execution/subagent-modes-detail.md));
+[`subagent-modes-detail` § Mode 8](../../contexts/execution/subagent-modes-detail.md));
 each prompt cites all four taxonomy statuses — see
 [`prompts/README.md`](prompts/README.md).
 
@@ -340,7 +340,7 @@ After every auto-dispatched run, write one audit-log-v1 line
 (`input_kind: "orchestration"`) via the `orchestration_record` recorder —
 never hand-author the JSON. Line shape, field semantics, recorder
 invocation, and `token_delta` sourcing priority →
-[`orchestration-telemetry` § Emit procedure](../../agent-src/contexts/execution/orchestration-telemetry.md).
+[`orchestration-telemetry` § Emit procedure](../../contexts/execution/orchestration-telemetry.md).
 Skip emit when `emergency.orchestration_halt` is set or `spawn_count == 0`
 (in-session run).
 
@@ -379,7 +379,7 @@ Skip emit when `emergency.orchestration_halt` is set or `spawn_count == 0`
 
 | Task                                 | Skill / command                      |
 |--------------------------------------|--------------------------------------|
-| Configuration reference              | [`subagent-configuration`](../../agent-src/contexts/execution/subagent-configuration.md) |
+| Configuration reference              | [`subagent-configuration`](../../contexts/subagent-configuration.md) |
 | Do-and-judge loop                    | Inline — see [`prompts/do-and-judge.md`](prompts/do-and-judge.md) |
 | Stepwise plan with judge gates       | Inline — see [`prompts/do-in-steps.md`](prompts/do-in-steps.md) |
 | Standalone judge on an existing diff | Inline — see judge prompt in [`prompts/do-and-judge.md`](prompts/do-and-judge.md) |

package/dist/agent-src/skills/subagent-orchestration/prompts/README.md

@@ -93,9 +93,10 @@ child continues the parent's task under identical tools and constraints (a
 fork inherits the parent's system prompt, tools, and conversation history
 exactly, so its first request reads the parent's cache). Prefer a **named
 subagent** when the child needs isolation, a different tool set, or nested
-dispatch — a fork cannot nest, and forking forces background mode, which
-changes the tool set and therefore invalidates the very prefix that
-motivated the fork in the first place. Never default to "always fork": that
+dispatch — a fork cannot nest. Forking forces background mode, which does
+not cost the prefix where the host row says `fork_shares_cache: true` (on
+Claude Code 2.1.302 a background fork read 99.8 % of its first call from the
+parent's cache and kept the parent's tool pool); elsewhere it is unobserved. Never default to "always fork": that
 option was considered and cut ([`../SKILL.md`](../SKILL.md) § Form gate)
 precisely because the cache-sharing benefit cannot be predicted before the
 fork happens, and background mode can erase it outright.

package/dist/agent-src/skills/svg-authoring/SKILL.md

@@ -174,7 +174,11 @@ nobody is still trying to match. Stop reporting it as progress and say why.
    traced silhouette that clips everything, then named regions split along an
    edge map; smooth, resample and simplify each contour. The outer edge comes
    from one contour and the holes come from the source. That trace is the
-   underlay; the shapes you ship are constructed on top of it.
+   underlay; the shapes you ship are constructed on top of it. **A shape with
+   counters** (o, e, a, B, 8, an O-ring mark) declares its `fill-rule`, winds
+   its outer and inner contours in opposite directions, and runs overlap removal
+   only after both are settled — two same-direction contours under the default
+   `nonzero` paint the counter solid.
 3. **Fit colour, do not pick it.** Least-squares per region against that region's
    own pixels; swap a linear gradient for a radial one only when the residual
    actually drops.
@@ -292,6 +296,8 @@ report a non-zero natural width.
   for the glyph it lacks.
 - `logo-generation` — the brand pipeline that
   routes a geometric mark here.
+- `type-design` — "the rest of the alphabet" / "make this a font": a
+  wordmark extended into a constructed font, reusing the trace and counter rule.
 - `image-provider-routing` — illustrative
   vector art goes to a provider, not here.
 - `fe-design` § Motion — the motion authority, in

package/dist/agent-src/skills/systematic-debugging/SKILL.md

@@ -278,7 +278,7 @@ If the root cause traces to a `agents/knowledge/` page that was
 followed while implementing (a documented convention was wrong, an
 API shape page was stale, a procedure was incomplete), append a
 `mistake_made` event to the knowledge intake — never rewrite the page
-mid-task (see [`knowledge-pages`](../../agent-src/templates/contexts/knowledge-pages.md)):
+mid-task (see [`knowledge-pages`](../../templates/contexts/knowledge-pages.md)):
 
 ```bash
 npx tsx node_modules/@event4u/agent-config/src/scripts/emit_knowledge_event.ts \
@@ -297,7 +297,7 @@ finds the new line.
 wrong right now (observed reality ≠ documented claim, not a one-off),
 this is the hybrid immediate-fix case instead — surface the proposed
 correction and ask before continuing (see
-[`knowledge-pages`](../../agent-src/templates/contexts/knowledge-pages.md)
+[`knowledge-pages`](../../templates/contexts/knowledge-pages.md)
 § Contested entries). Approved → isolated fix commit. Declined →
 `npx tsx node_modules/@event4u/agent-config/src/scripts/append_contested.ts` on that page, then still
 emit the `context_stale` event above it for the consolidation pass.

package/dist/agent-src/skills/tailwind-engineer/SKILL.md

@@ -201,8 +201,8 @@ Risks:          <arbitrary values, !important, dark-mode gaps>
   they read as auto-generated (catalog C1/C2). Reach for a single accent
   from the token map, or a duotone built from configured tokens.
 - **Anti-AI-slop: typography.** Unless audit-pinned, avoid surfacing
-  the system stack (`font-sans` fallback to Arial / Helvetica / Inter
-  via system defaults) as the *visible* body face. If `tailwind.config`
+  the default as the *visible* body face — Inter as a template webfont,
+  or `font-sans` (Arial / Helvetica). If `tailwind.config`
   pins a font family, use it; if not, treat the missing token as a
   gap to flag, not a license to ship the OS default.
 - **Anti-AI-slop: layout.** Unless audit-pinned, the centered hero +
@@ -217,7 +217,12 @@ When `DESIGN.md` declares `## Taste Dials`, honour them: Variance → layout-fam
 
 ## Security constraints
 
-`scripts/tailwind_config_gen.ts` is the only shipped script.
+`scripts/tailwind_config_gen.ts` is the only shipped script, started as:
+
+```bash
+node <skills-root>/tailwind-engineer/scripts/tailwind_config_gen.mjs \
+  [--framework react|vue|svelte|nextjs] [--output PATH] [--validate-only]
+```
 
 - **What it may touch** — exactly one file: `--output PATH`, or
   `tailwind.config.ts` / `.js` in the current working directory when no