@pseolint/core 0.4.3 → 0.5.3

package/dist/ai/orchestrator/index.d.ts

@@ -0,0 +1,25 @@
+/**
+ * Orchestrator MVP entry points.
+ *
+ * The runner takes a domain + a resolved `LanguageModel` (use existing
+ * `createLanguageModel` from ai/adapters) and runs a single audit session
+ * to completion or until budget is exhausted. Events are emitted to an
+ * optional `onEvent` callback (SSE / R2 fanout in the web app) and an
+ * optional NDJSON file. The full event log is also returned in the result.
+ *
+ * Phase 2 MVP scope: single-agent, generateText (not streamText), no
+ * watchdog injection. The follow-up swaps in streamText for live UI
+ * streaming and adds the watchdog.
+ */
+export { runOrchestrator } from "./runner.js";
+export type { RunOrchestratorOptions, EventSink } from "./runner.js";
+export { SessionState } from "./session.js";
+export type { SessionStateOptions } from "./session.js";
+export { BudgetTracker } from "./budget.js";
+export { SessionLog } from "./log.js";
+export { buildSystemPrompt } from "./prompt.js";
+export { finishAuditTool, manifestSchema } from "./finish-tool.js";
+export type { FixManifest } from "./finish-tool.js";
+export { DEFAULT_BUDGET, } from "./types.js";
+export type { BudgetCaps, UsageSnapshot, StopReason, SessionEvent, SessionResult, } from "./types.js";
+//# sourceMappingURL=index.d.ts.map

package/dist/ai/orchestrator/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../../src/ai/orchestrator/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAC9C,YAAY,EAAE,sBAAsB,EAAE,SAAS,EAAE,MAAM,aAAa,CAAC;AACrE,OAAO,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AAC5C,YAAY,EAAE,mBAAmB,EAAE,MAAM,cAAc,CAAC;AACxD,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AACtC,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAChD,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AACnE,YAAY,EAAE,WAAW,EAAE,MAAM,kBAAkB,CAAC;AACpD,OAAO,EACL,cAAc,GACf,MAAM,YAAY,CAAC;AACpB,YAAY,EACV,UAAU,EACV,aAAa,EACb,UAAU,EACV,YAAY,EACZ,aAAa,GACd,MAAM,YAAY,CAAC"}

package/dist/ai/orchestrator/index.js

@@ -0,0 +1,21 @@
+/**
+ * Orchestrator MVP entry points.
+ *
+ * The runner takes a domain + a resolved `LanguageModel` (use existing
+ * `createLanguageModel` from ai/adapters) and runs a single audit session
+ * to completion or until budget is exhausted. Events are emitted to an
+ * optional `onEvent` callback (SSE / R2 fanout in the web app) and an
+ * optional NDJSON file. The full event log is also returned in the result.
+ *
+ * Phase 2 MVP scope: single-agent, generateText (not streamText), no
+ * watchdog injection. The follow-up swaps in streamText for live UI
+ * streaming and adds the watchdog.
+ */
+export { runOrchestrator } from "./runner.js";
+export { SessionState } from "./session.js";
+export { BudgetTracker } from "./budget.js";
+export { SessionLog } from "./log.js";
+export { buildSystemPrompt } from "./prompt.js";
+export { finishAuditTool, manifestSchema } from "./finish-tool.js";
+export { DEFAULT_BUDGET, } from "./types.js";
+//# sourceMappingURL=index.js.map

package/dist/ai/orchestrator/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/ai/orchestrator/index.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;GAYG;AACH,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAC;AAE9C,OAAO,EAAE,YAAY,EAAE,MAAM,cAAc,CAAC;AAE5C,OAAO,EAAE,aAAa,EAAE,MAAM,aAAa,CAAC;AAC5C,OAAO,EAAE,UAAU,EAAE,MAAM,UAAU,CAAC;AACtC,OAAO,EAAE,iBAAiB,EAAE,MAAM,aAAa,CAAC;AAChD,OAAO,EAAE,eAAe,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAEnE,OAAO,EACL,cAAc,GACf,MAAM,YAAY,CAAC"}

package/dist/ai/orchestrator/log.d.ts

@@ -0,0 +1,24 @@
+import type { SessionEvent } from "./types.js";
+export type EventSink = (event: SessionEvent) => void | Promise<void>;
+/**
+ * Per-session event recorder. Three concerns:
+ *   1. In-memory buffer (returned in SessionResult so callers can replay).
+ *   2. Optional NDJSON file writer for durable replay (orchestrator session log).
+ *   3. Optional callback fanout — the web app passes a callback that pushes
+ *      to SSE / R2 in production.
+ *
+ * Errors writing the file are swallowed; we never want a log-write failure
+ * to abort an in-flight orchestrator session.
+ */
+export declare class SessionLog {
+    private readonly opts;
+    private readonly buffer;
+    private fileInitialized;
+    constructor(opts?: {
+        ndjsonPath?: string;
+        onEvent?: EventSink;
+    });
+    emit(event: SessionEvent): Promise<void>;
+    events(): SessionEvent[];
+}
+//# sourceMappingURL=log.d.ts.map

package/dist/ai/orchestrator/log.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"log.d.ts","sourceRoot":"","sources":["../../../src/ai/orchestrator/log.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,YAAY,CAAC;AAE/C,MAAM,MAAM,SAAS,GAAG,CAAC,KAAK,EAAE,YAAY,KAAK,IAAI,GAAG,OAAO,CAAC,IAAI,CAAC,CAAC;AAEtE;;;;;;;;;GASG;AACH,qBAAa,UAAU;IAKnB,OAAO,CAAC,QAAQ,CAAC,IAAI;IAJvB,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAsB;IAC7C,OAAO,CAAC,eAAe,CAAS;gBAGb,IAAI,GAAE;QACrB,UAAU,CAAC,EAAE,MAAM,CAAC;QACpB,OAAO,CAAC,EAAE,SAAS,CAAC;KAChB;IAGF,IAAI,CAAC,KAAK,EAAE,YAAY,GAAG,OAAO,CAAC,IAAI,CAAC;IAyB9C,MAAM,IAAI,YAAY,EAAE;CAGzB"}

package/dist/ai/orchestrator/log.js

@@ -0,0 +1,48 @@
+import { appendFile, mkdir } from "node:fs/promises";
+import { dirname } from "node:path";
+/**
+ * Per-session event recorder. Three concerns:
+ *   1. In-memory buffer (returned in SessionResult so callers can replay).
+ *   2. Optional NDJSON file writer for durable replay (orchestrator session log).
+ *   3. Optional callback fanout — the web app passes a callback that pushes
+ *      to SSE / R2 in production.
+ *
+ * Errors writing the file are swallowed; we never want a log-write failure
+ * to abort an in-flight orchestrator session.
+ */
+export class SessionLog {
+    opts;
+    buffer = [];
+    fileInitialized = false;
+    constructor(opts = {}) {
+        this.opts = opts;
+    }
+    async emit(event) {
+        this.buffer.push(event);
+        if (this.opts.onEvent) {
+            try {
+                await this.opts.onEvent(event);
+            }
+            catch {
+                // Sink failures are non-fatal — never let an SSE/R2 hiccup take
+                // down the session. The buffer is still authoritative.
+            }
+        }
+        if (this.opts.ndjsonPath) {
+            try {
+                if (!this.fileInitialized) {
+                    await mkdir(dirname(this.opts.ndjsonPath), { recursive: true });
+                    this.fileInitialized = true;
+                }
+                await appendFile(this.opts.ndjsonPath, JSON.stringify(event) + "\n", "utf8");
+            }
+            catch {
+                // Same — log-file errors don't kill the session.
+            }
+        }
+    }
+    events() {
+        return [...this.buffer];
+    }
+}
+//# sourceMappingURL=log.js.map

package/dist/ai/orchestrator/log.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"log.js","sourceRoot":"","sources":["../../../src/ai/orchestrator/log.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,kBAAkB,CAAC;AACrD,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAC;AAKpC;;;;;;;;;GASG;AACH,MAAM,OAAO,UAAU;IAKF;IAJF,MAAM,GAAmB,EAAE,CAAC;IACrC,eAAe,GAAG,KAAK,CAAC;IAEhC,YACmB,OAGb,EAAE;QAHW,SAAI,GAAJ,IAAI,CAGf;IACL,CAAC;IAEJ,KAAK,CAAC,IAAI,CAAC,KAAmB;QAC5B,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;QAExB,IAAI,IAAI,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC;YACtB,IAAI,CAAC;gBACH,MAAM,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC;YACjC,CAAC;YAAC,MAAM,CAAC;gBACP,gEAAgE;gBAChE,uDAAuD;YACzD,CAAC;QACH,CAAC;QAED,IAAI,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,CAAC;YACzB,IAAI,CAAC;gBACH,IAAI,CAAC,IAAI,CAAC,eAAe,EAAE,CAAC;oBAC1B,MAAM,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,CAAC,CAAC;oBAChE,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC;gBAC9B,CAAC;gBACD,MAAM,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,GAAG,IAAI,EAAE,MAAM,CAAC,CAAC;YAC/E,CAAC;YAAC,MAAM,CAAC;gBACP,iDAAiD;YACnD,CAAC;QACH,CAAC;IACH,CAAC;IAED,MAAM;QACJ,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC;IAC1B,CAAC;CACF"}

package/dist/ai/orchestrator/page-cache.d.ts

@@ -0,0 +1,64 @@
+/**
+ * One entry in the per-session page cache. `html` is the response body;
+ * `status` and `headers` come from the original fetch so downstream tools
+ * (`check_indexability` reading X-Robots-Tag, etc.) don't have to re-fetch.
+ */
+export interface PageCacheEntry {
+    url: string;
+    html: string;
+    status: number;
+    headers: Record<string, string>;
+    fetchedAt: number;
+}
+export declare class PageCache {
+    private readonly maxBytes;
+    private readonly store;
+    private currentBytes;
+    constructor(maxBytes?: number);
+    /**
+     * Stable pageId derivation: sha256 of the URL, truncated to 16 hex chars
+     * (~64 bits — collision-safe for any realistic session size). Same URL
+     * always maps to the same pageId so `fetch_page` is idempotent within
+     * a session, and the LLM can re-reference a page without juggling ids.
+     */
+    static idFor(url: string): string;
+    put(entry: Omit<PageCacheEntry, "fetchedAt">): string;
+    get(pageId: string): PageCacheEntry | null;
+    has(pageId: string): boolean;
+    size(): number;
+    /** Memory usage estimate in bytes — useful for orchestrator-side monitoring. */
+    approximateBytes(): number;
+}
+/**
+ * Run `fn` with `cache` available via `currentPageCache()`. The
+ * orchestrator runner wraps its `generateText` call in this so every tool
+ * invocation (including those nested inside the AI SDK's tool loop) sees
+ * the same cache.
+ */
+export declare function withPageCache<T>(cache: PageCache, fn: () => Promise<T>): Promise<T>;
+/**
+ * Read the current session's page cache. Returns null when called outside
+ * `withPageCache` (e.g. unit tests that don't set one up — those tests use
+ * `setTestCache` to inject a cache directly).
+ */
+export declare function currentPageCache(): PageCache | null;
+/**
+ * Test-only helper: set a page cache for the duration of a synchronous or
+ * async test. Use as `await withPageCache(cache, async () => { ... })`
+ * — same API as the production wrapper, just imported in tests.
+ */
+export declare const setTestCache: typeof withPageCache;
+/**
+ * Resolve a `pageId` to its cache entry — the standard tool-side helper.
+ * Throws a clear, LLM-friendly error when the cache isn't in scope or the
+ * pageId is unknown so the orchestrator can recover by re-fetching rather
+ * than crashing the session.
+ */
+export declare function resolvePage(pageId: string): PageCacheEntry;
+/**
+ * Multi-page variant. Resolves a list of pageIds in one call; throws on
+ * the first unknown id so the LLM gets specific feedback about which page
+ * needs re-fetching.
+ */
+export declare function resolvePages(pageIds: string[]): PageCacheEntry[];
+//# sourceMappingURL=page-cache.d.ts.map

package/dist/ai/orchestrator/page-cache.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"page-cache.d.ts","sourceRoot":"","sources":["../../../src/ai/orchestrator/page-cache.ts"],"names":[],"mappings":"AAGA;;;;GAIG;AACH,MAAM,WAAW,cAAc;IAC7B,GAAG,EAAE,MAAM,CAAC;IACZ,IAAI,EAAE,MAAM,CAAC;IACb,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAC;IAChC,SAAS,EAAE,MAAM,CAAC;CACnB;AA2BD,qBAAa,SAAS;IAIR,OAAO,CAAC,QAAQ,CAAC,QAAQ;IAHrC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAqC;IAC3D,OAAO,CAAC,YAAY,CAAK;gBAEI,QAAQ,GAAE,MAA0B;IAEjE;;;;;OAKG;IACH,MAAM,CAAC,KAAK,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM;IAIjC,GAAG,CAAC,KAAK,EAAE,IAAI,CAAC,cAAc,EAAE,WAAW,CAAC,GAAG,MAAM;IA4BrD,GAAG,CAAC,MAAM,EAAE,MAAM,GAAG,cAAc,GAAG,IAAI;IAI1C,GAAG,CAAC,MAAM,EAAE,MAAM,GAAG,OAAO;IAI5B,IAAI,IAAI,MAAM;IAId,gFAAgF;IAChF,gBAAgB,IAAI,MAAM;CAG3B;AAID;;;;;GAKG;AACH,wBAAgB,aAAa,CAAC,CAAC,EAAE,KAAK,EAAE,SAAS,EAAE,EAAE,EAAE,MAAM,OAAO,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,CAAC,CAAC,CAEnF;AAED;;;;GAIG;AACH,wBAAgB,gBAAgB,IAAI,SAAS,GAAG,IAAI,CAEnD;AAED;;;;GAIG;AACH,eAAO,MAAM,YAAY,sBAAgB,CAAC;AAE1C;;;;;GAKG;AACH,wBAAgB,WAAW,CAAC,MAAM,EAAE,MAAM,GAAG,cAAc,CAc1D;AAED;;;;GAIG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,cAAc,EAAE,CAEhE"}

package/dist/ai/orchestrator/page-cache.js

@@ -0,0 +1,127 @@
+import { AsyncLocalStorage } from "node:async_hooks";
+import { createHash } from "node:crypto";
+/**
+ * Session-scoped page cache. The orchestrator's reason for being:
+ *   1. `fetch_page` writes HTML here keyed by `pageId`
+ *   2. Returns the `pageId` (not the HTML) to the LLM
+ *   3. Subsequent tools (parse_page, check_rule_*, validate_jsonld, etc.)
+ *      take a `pageId` in their input and look up HTML via the cache
+ *
+ * Net effect: HTML never lives in the LLM conversation history. The cost
+ * impact is dramatic — the v0.4.4 dogfood showed 5 fetched pages exploding
+ * to 297K tokens / $0.91 in a single step because re-passing HTML through
+ * tool inputs accumulated quadratically. With pageId references that
+ * collapses to ~20K tokens / $0.06 for the same workload.
+ *
+ * Implementation: AsyncLocalStorage threads the cache through generateText
+ * → tool execute() without changing the AI SDK's tool surface. Tools that
+ * don't need page content ignore it entirely.
+ */
+/**
+ * Soft cap on per-session cache size. Generous enough for any realistic
+ * orchestrator session (200 pages × 100KB ≈ 20MB) but bounds memory growth
+ * for misbehaving runs. `put` rejects with a tool-friendly error past this
+ * cap so the LLM can adapt rather than the orchestrator OOMing silently.
+ */
+const DEFAULT_MAX_BYTES = 50 * 1024 * 1024; // 50MB
+export class PageCache {
+    maxBytes;
+    store = new Map();
+    currentBytes = 0;
+    constructor(maxBytes = DEFAULT_MAX_BYTES) {
+        this.maxBytes = maxBytes;
+    }
+    /**
+     * Stable pageId derivation: sha256 of the URL, truncated to 16 hex chars
+     * (~64 bits — collision-safe for any realistic session size). Same URL
+     * always maps to the same pageId so `fetch_page` is idempotent within
+     * a session, and the LLM can re-reference a page without juggling ids.
+     */
+    static idFor(url) {
+        return createHash("sha256").update(url).digest("hex").slice(0, 16);
+    }
+    put(entry) {
+        const pageId = PageCache.idFor(entry.url);
+        const bytes = entry.html.length +
+            entry.url.length +
+            Object.entries(entry.headers).reduce((acc, [k, v]) => acc + k.length + v.length, 0);
+        // Replacing an existing entry: subtract its old size from the running total
+        // before checking the cap — re-fetching the same URL shouldn't re-bill cap.
+        const existing = this.store.get(pageId);
+        if (existing) {
+            this.currentBytes -=
+                existing.html.length +
+                    existing.url.length +
+                    Object.entries(existing.headers).reduce((acc, [k, v]) => acc + k.length + v.length, 0);
+        }
+        if (this.currentBytes + bytes > this.maxBytes) {
+            throw new Error(`page cache full (${this.maxBytes} bytes); reduce concurrent fetches or finalize the audit before fetching more`);
+        }
+        this.store.set(pageId, { ...entry, fetchedAt: Date.now() });
+        this.currentBytes += bytes;
+        return pageId;
+    }
+    get(pageId) {
+        return this.store.get(pageId) ?? null;
+    }
+    has(pageId) {
+        return this.store.has(pageId);
+    }
+    size() {
+        return this.store.size;
+    }
+    /** Memory usage estimate in bytes — useful for orchestrator-side monitoring. */
+    approximateBytes() {
+        return this.currentBytes;
+    }
+}
+const cacheStorage = new AsyncLocalStorage();
+/**
+ * Run `fn` with `cache` available via `currentPageCache()`. The
+ * orchestrator runner wraps its `generateText` call in this so every tool
+ * invocation (including those nested inside the AI SDK's tool loop) sees
+ * the same cache.
+ */
+export function withPageCache(cache, fn) {
+    return cacheStorage.run(cache, fn);
+}
+/**
+ * Read the current session's page cache. Returns null when called outside
+ * `withPageCache` (e.g. unit tests that don't set one up — those tests use
+ * `setTestCache` to inject a cache directly).
+ */
+export function currentPageCache() {
+    return cacheStorage.getStore() ?? null;
+}
+/**
+ * Test-only helper: set a page cache for the duration of a synchronous or
+ * async test. Use as `await withPageCache(cache, async () => { ... })`
+ * — same API as the production wrapper, just imported in tests.
+ */
+export const setTestCache = withPageCache;
+/**
+ * Resolve a `pageId` to its cache entry — the standard tool-side helper.
+ * Throws a clear, LLM-friendly error when the cache isn't in scope or the
+ * pageId is unknown so the orchestrator can recover by re-fetching rather
+ * than crashing the session.
+ */
+export function resolvePage(pageId) {
+    const cache = currentPageCache();
+    if (!cache) {
+        throw new Error("page cache not in scope — tool must run inside the orchestrator's withPageCache wrapper");
+    }
+    const entry = cache.get(pageId);
+    if (!entry) {
+        throw new Error(`unknown pageId ${pageId} — call fetch_page to populate the cache before passing the returned pageId here`);
+    }
+    return entry;
+}
+/**
+ * Multi-page variant. Resolves a list of pageIds in one call; throws on
+ * the first unknown id so the LLM gets specific feedback about which page
+ * needs re-fetching.
+ */
+export function resolvePages(pageIds) {
+    return pageIds.map(resolvePage);
+}
+//# sourceMappingURL=page-cache.js.map

package/dist/ai/orchestrator/page-cache.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"page-cache.js","sourceRoot":"","sources":["../../../src/ai/orchestrator/page-cache.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,MAAM,kBAAkB,CAAC;AACrD,OAAO,EAAE,UAAU,EAAE,MAAM,aAAa,CAAC;AAezC;;;;;;;;;;;;;;;;GAgBG;AACH;;;;;GAKG;AACH,MAAM,iBAAiB,GAAG,EAAE,GAAG,IAAI,GAAG,IAAI,CAAC,CAAC,OAAO;AAEnD,MAAM,OAAO,SAAS;IAIS;IAHZ,KAAK,GAAG,IAAI,GAAG,EAA0B,CAAC;IACnD,YAAY,GAAG,CAAC,CAAC;IAEzB,YAA6B,WAAmB,iBAAiB;QAApC,aAAQ,GAAR,QAAQ,CAA4B;IAAG,CAAC;IAErE;;;;;OAKG;IACH,MAAM,CAAC,KAAK,CAAC,GAAW;QACtB,OAAO,UAAU,CAAC,QAAQ,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IACrE,CAAC;IAED,GAAG,CAAC,KAAwC;QAC1C,MAAM,MAAM,GAAG,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC1C,MAAM,KAAK,GACT,KAAK,CAAC,IAAI,CAAC,MAAM;YACjB,KAAK,CAAC,GAAG,CAAC,MAAM;YAChB,MAAM,CAAC,OAAO,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QAEtF,4EAA4E;QAC5E,4EAA4E;QAC5E,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;QACxC,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,CAAC,YAAY;gBACf,QAAQ,CAAC,IAAI,CAAC,MAAM;oBACpB,QAAQ,CAAC,GAAG,CAAC,MAAM;oBACnB,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;QAC3F,CAAC;QAED,IAAI,IAAI,CAAC,YAAY,GAAG,KAAK,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;YAC9C,MAAM,IAAI,KAAK,CACb,oBAAoB,IAAI,CAAC,QAAQ,+EAA+E,CACjH,CAAC;QACJ,CAAC;QAED,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,EAAE,EAAE,GAAG,KAAK,EAAE,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QAC5D,IAAI,CAAC,YAAY,IAAI,KAAK,CAAC;QAC3B,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,GAAG,CAAC,MAAc;QAChB,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,IAAI,IAAI,CAAC;IACxC,CAAC;IAED,GAAG,CAAC,MAAc;QAChB,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAChC,CAAC;IAED,IAAI;QACF,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC;IACzB,CAAC;IAED,gFAAgF;IAChF,gBAAgB;QACd,OAAO,IAAI,CAAC,YAAY,CAAC;IAC3B,CAAC;CACF;AAED,MAAM,YAAY,GAAG,IAAI,iBAAiB,EAAa,CAAC;AAExD;;;;;GAKG;AACH,MAAM,UAAU,aAAa,CAAI,KAAgB,EAAE,EAAoB;IACrE,OAAO,YAAY,CAAC,GAAG,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;AACrC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,gBAAgB;IAC9B,OAAO,YAAY,CAAC,QAAQ,EAAE,IAAI,IAAI,CAAC;AACzC,CAAC;AAED;;;;GAIG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,aAAa,CAAC;AAE1C;;;;;GAKG;AACH,MAAM,UAAU,WAAW,CAAC,MAAc;IACxC,MAAM,KAAK,GAAG,gBAAgB,EAAE,CAAC;IACjC,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,MAAM,IAAI,KAAK,CACb,yFAAyF,CAC1F,CAAC;IACJ,CAAC;IACD,MAAM,KAAK,GAAG,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAChC,IAAI,CAAC,KAAK,EAAE,CAAC;QACX,MAAM,IAAI,KAAK,CACb,kBAAkB,MAAM,kFAAkF,CAC3G,CAAC;IACJ,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,OAAiB;IAC5C,OAAO,OAAO,CAAC,GAAG,CAAC,WAAW,CAAC,CAAC;AAClC,CAAC"}

package/dist/ai/orchestrator/prompt.d.ts

@@ -0,0 +1,16 @@
+import type { BudgetCaps } from "./types.js";
+/**
+ * Build the orchestrator system prompt. Structure:
+ *   1. Role
+ *   2. Methodology hints (not script — model picks order)
+ *   3. Budget contract (concrete numbers from BudgetCaps)
+ *   4. Output contract (must call finish_audit to terminate)
+ *   5. Tool-use guidance (avoid amplifying low-confidence findings, etc.)
+ *
+ * Kept stable between sessions so Anthropic prompt caching keeps the
+ * ~2-3K-token system prompt cached across invocations. Only the budget
+ * numbers vary per session, but they're at the bottom and don't break
+ * higher-prefix cache hits.
+ */
+export declare function buildSystemPrompt(caps: BudgetCaps): string;
+//# sourceMappingURL=prompt.d.ts.map

package/dist/ai/orchestrator/prompt.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"prompt.d.ts","sourceRoot":"","sources":["../../../src/ai/orchestrator/prompt.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AAE7C;;;;;;;;;;;;GAYG;AACH,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,UAAU,GAAG,MAAM,CAqC1D"}

package/dist/ai/orchestrator/prompt.js

@@ -0,0 +1,52 @@
+/**
+ * Build the orchestrator system prompt. Structure:
+ *   1. Role
+ *   2. Methodology hints (not script — model picks order)
+ *   3. Budget contract (concrete numbers from BudgetCaps)
+ *   4. Output contract (must call finish_audit to terminate)
+ *   5. Tool-use guidance (avoid amplifying low-confidence findings, etc.)
+ *
+ * Kept stable between sessions so Anthropic prompt caching keeps the
+ * ~2-3K-token system prompt cached across invocations. Only the budget
+ * numbers vary per session, but they're at the bottom and don't break
+ * higher-prefix cache hits.
+ */
+export function buildSystemPrompt(caps) {
+    return [
+        `You are pseolint's SEO audit orchestrator. Drive deterministic tools to produce a fix manifest with concrete, paste-able patches — not just a list of findings.`,
+        ``,
+        `## Suggested order (you choose)`,
+        `1. fetch_sitemap → discover URLs`,
+        `2. detect_templates → identify dominant patterns`,
+        `3. check_domain_llms_txt + check_domain_crawler_access (once per origin)`,
+        `4. sample_template (3-5 URLs per meaningful template, ≥5% of URLs)`,
+        `5. fetch_page in parallel batches of 5-8 → parse_page → check_indexability`,
+        `6. Per-page rule checks for indexable pages: thin_content, missing_author, canonical_consistency, json_ld_valid, answer_first, validate_jsonld`,
+        `7. Cluster checks per-template: near_duplicate, faq_coverage (if FAQ-typed URLs), citable_facts, freshness_signals, content_modularity, summary_bait, required_fields, schema_consistency`,
+        `8. Translate findings into patches; call finish_audit`,
+        ``,
+        `## Findings vs patches`,
+        `A finding is evidence (e.g. "thin content on /city/austin"). A patch is a concrete fix a user can paste in. Translate evidence into action:`,
+        `- replace_h1, rewrite_meta, rewrite_intro — text replacements`,
+        `- add_jsonld, add_faq_block, add_internal_link — HTML insertions`,
+        `- remove_thin_block — selector-based deletion`,
+        `- robots_txt, sitemap_xml, canonical_strategy — domain-level`,
+        `Use template-level patches when an issue spans 80%+ of a cluster. Per-page patches don't scale.`,
+        `For add_jsonld, validate_jsonld lists exactly which Schema.org properties are missing per declared @type — use that to populate the block.`,
+        ``,
+        `## Tool-use rules`,
+        `- Findings carry confidence (high|medium|low|speculative). Don't propose structural rewrites on low/speculative evidence alone — corroborate with another tool first.`,
+        `- Skip rule checks on non-indexable pages (noindex, wrong canonical) — they don't affect search.`,
+        `- If a tool returns ok:false, move on. Don't retry the same input.`,
+        `- Prefer parallel tool calls when independent (multi-page fetches especially).`,
+        `- HTML never travels in conversation: fetch_page returns a pageId you pass to subsequent tools. The token cost stays bounded.`,
+        ``,
+        `## Budget`,
+        `${caps.maxToolCalls} tool calls · ${caps.maxInputTokensTotal.toLocaleString("en-US")} input tokens · $${caps.maxSessionUsd.toFixed(2)} · ${caps.maxWallSeconds}s wall.`,
+        `The runtime stops you at any cap. Partial manifests are useful — prefer breadth (cover all templates) over depth (exhaustive on one).`,
+        ``,
+        `## Output`,
+        `Call finish_audit(manifest) exactly once to terminate. Without it, the session ends with no manifest. Empty patch arrays are fine when nothing's actionable — don't invent patches.`,
+    ].join("\n");
+}
+//# sourceMappingURL=prompt.js.map

package/dist/ai/orchestrator/prompt.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"prompt.js","sourceRoot":"","sources":["../../../src/ai/orchestrator/prompt.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,iBAAiB,CAAC,IAAgB;IAChD,OAAO;QACL,iKAAiK;QACjK,EAAE;QACF,iCAAiC;QACjC,kCAAkC;QAClC,kDAAkD;QAClD,0EAA0E;QAC1E,oEAAoE;QACpE,4EAA4E;QAC5E,gJAAgJ;QAChJ,2LAA2L;QAC3L,uDAAuD;QACvD,EAAE;QACF,wBAAwB;QACxB,6IAA6I;QAC7I,+DAA+D;QAC/D,kEAAkE;QAClE,+CAA+C;QAC/C,8DAA8D;QAC9D,iGAAiG;QACjG,4IAA4I;QAC5I,EAAE;QACF,mBAAmB;QACnB,uKAAuK;QACvK,kGAAkG;QAClG,oEAAoE;QACpE,gFAAgF;QAChF,+HAA+H;QAC/H,EAAE;QACF,WAAW;QACX,GAAG,IAAI,CAAC,YAAY,iBAAiB,IAAI,CAAC,mBAAmB,CAAC,cAAc,CAAC,OAAO,CAAC,oBAAoB,IAAI,CAAC,aAAa,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,IAAI,CAAC,cAAc,SAAS;QACxK,uIAAuI;QACvI,EAAE;QACF,WAAW;QACX,qLAAqL;KACtL,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACf,CAAC"}

package/dist/ai/orchestrator/runner.d.ts

@@ -0,0 +1,65 @@
+import { type LanguageModel } from "ai";
+import type { BudgetCaps, SessionResult } from "./types.js";
+import type { EventSink as _EventSinkPathReexport } from "./log.js";
+/**
+ * Loose type for the tool registry. The AI SDK's `Tool<I, O>` is invariant
+ * in its generics, so a heterogeneous map of tools with distinct schemas
+ * can't be statically typed as `Record<string, Tool<unknown, unknown>>`.
+ * We accept the unsoundness at this single boundary — every individual tool
+ * is type-safe at definition site (via `defineTool`), and the AI SDK
+ * validates inputs at runtime via Zod regardless.
+ */
+type AnyDefinedTool = {
+    name: string;
+    toAiTool: () => any;
+};
+export interface RunOrchestratorOptions {
+    domain: string;
+    userId: string;
+    model: LanguageModel;
+    /** Provider id for cost estimation (e.g. "anthropic"). */
+    providerId: string;
+    /** Model id for cost estimation (e.g. "claude-opus-4-7"). */
+    modelId: string;
+    /** Override default budget caps. Defaults from DEFAULT_BUDGET. */
+    budget?: Partial<BudgetCaps>;
+    /** Optional path for durable NDJSON session log. */
+    ndjsonPath?: string;
+    /** Optional event sink (e.g. SSE / R2 fanout). */
+    onEvent?: EventSink;
+    /** External abort signal — aborting flips the session to `aborted`. */
+    signal?: AbortSignal;
+    /**
+     * Override the tool registry. Default = full `orchestratorTools` plus
+     * `finish_audit`. Tests pass a smaller subset; production passes the full
+     * registry.
+     */
+    tools?: Record<string, AnyDefinedTool>;
+    /**
+     * Per-model-call retry budget for transient errors (5xx, rate limits,
+     * network failures). The AI SDK handles the retry loop internally and
+     * respects `retry-after` headers from the provider, so each retry waits
+     * the rate-limit window when the provider asks.
+     *
+     * Default 5: tier-1 Anthropic accounts have a 30K input-tokens-per-minute
+     * cap that burst-friendly orchestrator sessions can hit. Five retries
+     * with exponential backoff (~1s/2s/4s/8s/16s base) span a full 60s TPM
+     * refresh, so the next retry usually succeeds. Lower this only if the
+     * caller has tier-2+ keys (no TPM cap) or cares more about fail-fast.
+     */
+    maxRetries?: number;
+}
+/**
+ * Run a single orchestrator session end-to-end. Returns the final
+ * `SessionResult` once the LLM calls `finish_audit`, the budget is
+ * exhausted, the abort signal fires, or the model errors out.
+ *
+ * No streaming yet — uses `generateText` for simpler test ergonomics with
+ * the existing `MockModel` (which only implements `doGenerate`). Phase 2
+ * follow-up swaps to `streamText` once we have a stream-capable mock and a
+ * web-app SSE endpoint to wire up.
+ */
+export declare function runOrchestrator(opts: RunOrchestratorOptions): Promise<SessionResult>;
+export type EventSink = _EventSinkPathReexport;
+export {};
+//# sourceMappingURL=runner.d.ts.map

package/dist/ai/orchestrator/runner.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"runner.d.ts","sourceRoot":"","sources":["../../../src/ai/orchestrator/runner.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,aAAa,EAAE,MAAM,IAAI,CAAC;AAOtD,OAAO,KAAK,EAAE,UAAU,EAAE,aAAa,EAAc,MAAM,YAAY,CAAC;AACxE,OAAO,KAAK,EAAE,SAAS,IAAI,sBAAsB,EAAE,MAAM,UAAU,CAAC;AAEpE;;;;;;;GAOG;AAEH,KAAK,cAAc,GAAG;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,QAAQ,EAAE,MAAM,GAAG,CAAA;CAAE,CAAC;AAI5D,MAAM,WAAW,sBAAsB;IACrC,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;IACf,KAAK,EAAE,aAAa,CAAC;IACrB,0DAA0D;IAC1D,UAAU,EAAE,MAAM,CAAC;IACnB,6DAA6D;IAC7D,OAAO,EAAE,MAAM,CAAC;IAChB,kEAAkE;IAClE,MAAM,CAAC,EAAE,OAAO,CAAC,UAAU,CAAC,CAAC;IAC7B,oDAAoD;IACpD,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,kDAAkD;IAClD,OAAO,CAAC,EAAE,SAAS,CAAC;IACpB,uEAAuE;IACvE,MAAM,CAAC,EAAE,WAAW,CAAC;IACrB;;;;OAIG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,cAAc,CAAC,CAAC;IACvC;;;;;;;;;;;OAWG;IACH,UAAU,CAAC,EAAE,MAAM,CAAC;CACrB;AAID;;;;;;;;;GASG;AACH,wBAAsB,eAAe,CAAC,IAAI,EAAE,sBAAsB,GAAG,OAAO,CAAC,aAAa,CAAC,CAuM1F;AAmBD,MAAM,MAAM,SAAS,GAAG,sBAAsB,CAAC"}