feat(ts): add real quality gates — Biome lint + effect-law ratchet + class inventory

- biome.json (2.4.16, linter-only) wired as "lint" in all six packages
- scripts/check-effect-laws.ts: Effect-native law enforcement encoding the
  adapted beep-effect effect-first/schema-first laws (no native JSON/switch/
  sort/fetch/timers, no process.env, no throw new, no Effect.run* outside
  boundaries, no Schema-suffixed constants, no node:fs/path, AST-based
  pure-data interface detection per law 38/39)
- ratcheting baseline allowlist (95 entries / 290 findings) that must shrink
  to documented exemptions only; stale counts fail the gate
- root lint chains turbo lint + law check + native-class inventory
- fix all 163 initial Biome findings: import-type style, templates, two `any`s,
  ten non-null assertions (librarian getService gate, A.matchRight in atoms,
  ensureNode returning nodes, main.tsx mount guard)

Gates: lint, check:tsgo, build, test (force, 11 tasks) all green.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

ts/packages/flow/src/model/text-completion/azure-openai.ts

@@ -10,16 +10,22 @@
 
 import { AzureOpenAI } from "openai";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+  LlmResult,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
-  type LlmResult,
 } from "@trustgraph/base";
 import { Effect, Stream } from "effect";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   llmStreamPart,
   makeTextCompletionLayer,
@@ -27,8 +33,6 @@ import {
   providerStatusError,
   requiredString,
   streamTextCompletionChunks,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type AzureOpenAIProcessorConfig = ProcessorConfig & {

ts/packages/flow/src/model/text-completion/claude.ts

@@ -6,23 +6,27 @@
 
 import { AnthropicClient, AnthropicLanguageModel } from "@effect/ai-anthropic";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
 } from "@trustgraph/base";
 import { Effect, Layer, Redacted } from "effect";
 import { FetchHttpClient } from "effect/unstable/http";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   makeLanguageModelProvider,
   makeTextCompletionLayer,
   optionalStringConfig,
   requiredString,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type ClaudeProcessorConfig = ProcessorConfig & {

ts/packages/flow/src/model/text-completion/common.ts

@@ -1,18 +1,22 @@
+import type {
+  LlmChunk,
+  LlmResult,
+  LlmProvider,
+} from "@trustgraph/base";
 import {
   Llm,
   TooManyRequestsError,
   errorMessage,
   makeLlmServiceShape,
-  type LlmChunk,
-  type LlmResult,
-  type LlmProvider,
 } from "@trustgraph/base";
-import { Config, Context, Effect, Layer, Match, Ref, Result, Stream } from "effect";
+import type { Context, } from "effect";
+import { Config, Effect, Layer, Match, Ref, Result, Stream } from "effect";
 import * as O from "effect/Option";
 import * as Predicate from "effect/Predicate";
 import * as S from "effect/Schema";
 import type * as Scope from "effect/Scope";
-import { AiError, LanguageModel, Prompt, Response } from "effect/unstable/ai";
+import type { LanguageModel, Prompt, Response } from "effect/unstable/ai";
+import { AiError, } from "effect/unstable/ai";
 
 export class TextCompletionConfigError extends S.TaggedErrorClass<TextCompletionConfigError>()(
   "TextCompletionConfigError",

ts/packages/flow/src/model/text-completion/mistral.ts

@@ -8,16 +8,22 @@
 
 import { Mistral } from "@mistralai/mistralai";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+  LlmResult,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
-  type LlmResult,
 } from "@trustgraph/base";
 import { Effect, Stream } from "effect";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   llmStreamPart,
   makeTextCompletionLayer,
@@ -26,8 +32,6 @@ import {
   requiredString,
   streamTextCompletionChunks,
   textFromContent,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type MistralProcessorConfig = ProcessorConfig & {

ts/packages/flow/src/model/text-completion/ollama.ts

@@ -8,24 +8,28 @@
 
 import { Ollama } from "ollama";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+  LlmResult,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
-  type LlmResult,
 } from "@trustgraph/base";
 import { Effect, Stream } from "effect";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   llmStreamPart,
   makeTextCompletionLayer,
   optionalStringConfig,
   providerRuntimeError,
   streamTextCompletionChunks,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type OllamaProcessorConfig = ProcessorConfig & {
@@ -68,7 +72,7 @@ const makeOllamaProviderFromClient = (
       _temperature?: number,
     ) => {
       const modelName = model ?? defaultModel;
-      const fullPrompt = system + "\n\n" + prompt;
+      const fullPrompt = `${system}\n\n${prompt}`;
 
       return Effect.tryPromise({
         try: () =>
@@ -95,7 +99,7 @@ const makeOllamaProviderFromClient = (
       _temperature?: number,
     ) => {
       const modelName = model ?? defaultModel;
-      const fullPrompt = system + "\n\n" + prompt;
+      const fullPrompt = `${system}\n\n${prompt}`;
 
       return Stream.fromEffect(
         Effect.tryPromise({

ts/packages/flow/src/model/text-completion/openai-compatible.ts

@@ -11,16 +11,22 @@
 
 import OpenAI from "openai";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+  LlmResult,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
-  type LlmResult,
 } from "@trustgraph/base";
 import { Effect, Stream } from "effect";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   llmStreamPart,
   makeTextCompletionLayer,
@@ -28,8 +34,6 @@ import {
   providerStatusError,
   requiredString,
   streamTextCompletionChunks,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type OpenAICompatibleProcessorConfig = ProcessorConfig & {

ts/packages/flow/src/model/text-completion/openai.ts

@@ -6,16 +6,22 @@
 
 import OpenAI from "openai";
 import { NodeRuntime } from "@effect/platform-node";
+import type {
+  Llm,
+  LlmProvider,
+  ProcessorConfig,
+  LlmResult,
+} from "@trustgraph/base";
 import {
   makeLlmService,
   makeFlowProcessorProgram,
   makeLlmSpecs,
-  type Llm,
-  type LlmProvider,
-  type ProcessorConfig,
-  type LlmResult,
 } from "@trustgraph/base";
 import { Effect, Stream } from "effect";
+import type {
+  TextCompletionConfigError,
+  TextCompletionRuntimeError,
+} from "./common.ts";
 import {
   llmStreamPart,
   makeTextCompletionLayer,
@@ -23,8 +29,6 @@ import {
   providerStatusError,
   requiredString,
   streamTextCompletionChunks,
-  type TextCompletionConfigError,
-  type TextCompletionRuntimeError,
 } from "./common.ts";
 
 export type OpenAIProcessorConfig = ProcessorConfig & {