@autorag/librarian 2.10.5 → 2.11.0

package/skills/autorag-setup/SKILL.md

@@ -21,7 +21,8 @@ librarian, not the Lite MCP server.
 - Never print, copy, migrate, compare, or persist credential values. Store only
   environment-variable names such as `apiKeyEnv`.
 - Do not scan the whole filesystem or home directory without explicit approval.
-- Never move, rename, edit, or delete source documents.
+- Setup, indexing, and refresh never move, rename, edit, or delete source
+  documents. The librarian itself may organize files when the user asks it to.
 - Do not index system trees, app bundles, caches, credential stores,
   `node_modules`, `.git`, `dist`, `build`, `target`, `.cache`, `.autorag`, or
   `.jikji`.
@@ -189,20 +190,28 @@ is impossible, refresh continues without this optimization and the user is
 told duplicate exclusion is off. Set `"excludeExactDuplicates": false` to
 index every copy.
 
-MinSync's default embedder is in-process native Qwen3 embeddings
-(`native:Qwen/Qwen3-Embedding-0.6B`, 1024 dimensions, MinSync 0.4.10). The
-default needs no embedder flags, no API key, and no external daemon (such as
-Ollama) — all embeddings run in-process locally and privately. For workspaces
-using the loopback llama-server gateway, prefetch the verified model with
-`autorag models prefetch --profile qwen3-embedding-0.6b` (or verify with
-`autorag models verify --profile qwen3-embedding-0.6b`).
-
-The legacy Ollama/TEI adapter path (EmbeddingGemma, 768 dimensions) is
-supported only for existing legacy workspaces and manual QA; do not use it as a
-fresh install default.
-
-Override the embedder only when intentionally using a different, for example
-remote, provider:
+AutoRAG owns no embedding model. MinSync provisions the embedder
+(`Qwen/Qwen3-Embedding-0.6B`, 1024 dimensions) in its own cache
+(`~/.cache/minsync/models`); there is no local model download and no bundled
+runtime. One shared embedder serves everything that embeds inside
+AutoRAG: MinSync indexing/query, memory similar-question search, the
+chat-crawler sidecars, the GitHub Gist vector index, and the endpoint external
+CLIs are pointed at. The default needs no embedder flags and no API key. It
+resolves in order:
+
+1. A configured remote endpoint (`minSync.embedder`), used for everything; no
+   daemon is started and no local model is loaded.
+2. A running `minsync serve` daemon (loopback only; MinSync 0.4.11+).
+   AutoRAG never starts it: run `minsync serve --idle-timeout 0` yourself, as
+   a launchd/systemd service (see `docs/minsync-setup.md`), or press
+   Ctrl+Shift+E in `autorag tui`. `autorag setup` only reports whether it is
+   alive.
+3. MinSync's in-process engine, through the optional Node-API addon
+   `@nomadamas/minsync-native` (same engine and weights, so bit-for-bit
+   identical vectors; not bundled). Without a daemon and without the addon,
+   embedding fails with an error naming the missing package.
+
+Override the embedder only when intentionally using a remote provider:
 
 ```bash
 autorag init \
@@ -213,9 +222,14 @@ autorag init \
   --embedder-batch-size 64
 ```
 
-Only store the environment-variable name, never its value. Dimension and batch
-size must be positive integers, and the dimension must match the embedder
-(default Qwen3 is 1024; legacy EmbeddingGemma is 768; `voyageai/voyage-4-lite` via OpenRouter is 1024).
+A remote `minSync.embedder` names an `id` (`openai:<model>`, `tei:<model>`, or
+any id together with a `baseUrl`) plus optional `apiKeyEnv` (the **name** of the
+environment variable holding the bearer token), `dimension`, `batchSize`, and
+`timeoutMs`. OpenAI-protocol endpoints POST `{baseUrl}/v1/embeddings`; a `tei:`
+endpoint POSTs `{baseUrl}/embed`; `openai:` with no `baseUrl` means
+`https://api.openai.com`. Only store the environment-variable name, never its
+value. `dimension` and `batchSize` must be positive integers, and `dimension`
+must match the endpoint (the local default is 1024).
 
 ### Jev routing and question decomposition (on by default)
 
@@ -336,7 +350,7 @@ keeps its mirrors and reports `search-root-unavailable`; see
 
 ## Probe and configure datasource skills (setup wizard)
 
-Use `autorag setup` to probe the local runtime, model profile, and known
+Use `autorag setup` to probe the embedding backend, MinSync, and known
 datasources automatically:
 
 ```bash
@@ -348,7 +362,7 @@ setup. Configure datasources directly in trusted config, wizard-style:
 
 1. Probe every datasource for setup feasibility before asking the user
    anything: the backing CLI exists (`lazykatok`, `discrawl`, `slacrawl`,
-   `wacrawl`, `telecrawl`, `notcrawl`, `qmd`, `mailcrawl`, `rclone`, `lark-cli`) and its
+   `wacrawl`, `telecrawl`, `notcrawl`, `mailcrawl`, `rclone`, `lark-cli`) and its
    local store or archive is present. CLI-backed datasources own their own
    archive, index, and authentication, so environment credentials (such as bot
    tokens) are never required or checked for them. Non-CLI connectors
@@ -371,7 +385,7 @@ setup. Configure datasources directly in trusted config, wizard-style:
 Datasource skills belong in trusted config. Builtin
 template names are `kakao`, `whatsapp`, `telegram`, `slack`, `discord`,
 `clawgallery`, `notion`, `github`, `github-gist`, `cloud-drive`, `mail-export`,
-`mailcrawl`, `obsidian`, `rss`, `spotlight`, and `lark`. Config keys may be connection
+`mailcrawl`, `obsidian`, `rss`, `spotlight`, `lark`, and `hister`. Config keys may be connection
 aliases with `"type": "<template>"`. Unknown names are skipped with an
 `unknown-datasource-skill` warning; they do not fail config resolution.
 `scope` narrows a query to a sub-path as ordinary filtering. Tags are descriptive metadata only, not search filters. MCP `datasourceIds` selects configured connections before retrieval; discover their IDs with `autorag.datasources.list`.
@@ -393,16 +407,74 @@ aliases with `"type": "<template>"`. Unknown names are skipped with an
 Tokens are environment-variable names, not raw secrets. CLI-backed connectors
 keep authentication in their external tool configuration.
 
-Mailcrawl must be installed separately (`@nomadamas/mailcrawl@0.2.0` or newer)
+Mailcrawl must be installed separately (`@nomadamas/mailcrawl@0.2.2` or newer)
 and configured through its own Himalaya account. AutoRAG runs its local `sync`
 and `index` lifecycle, then uses the mailcrawl CLI for BM25, semantic, or
 hybrid search. Do not use 0.1.3 or earlier: a no-op sync followed by `index`
-fails with `text array must be non-empty`. 0.2.0 defaults to the in-process
-native `Qwen/Qwen3-Embedding-0.6B` embedder and keeps vectors in LanceDB, so a
-cold cache makes the first `index` download ONNX weights and run for minutes.
+fails with `text array must be non-empty`. 0.2.2 embeds through the same
+backend as the rest of AutoRAG (a remote `minSync.embedder`, else the
+`minsync serve` daemon) instead of its own in-process model, and keeps vectors
+in LanceDB. Switching from the old native model rebuilds the semantic index on
+the next `index`. kakao needs `lazykatok` 0.3.3 or newer for the same reason.
 Use mailcrawl for Gmail, IMAP, and Maildir retrieval. The former Gmail REST
 datasource is removed.
 
+### Browser history (Hister) and the shared embedding model
+
+`hister` is the one datasource whose backing program AutoRAG does not install,
+start, or bundle (it is AGPL-3.0-or-later and stays a separate program). Setting
+it up is an agent task with these steps. Run each command; do not just describe it.
+
+1. **Probe first.** Add the datasource entry, then run `autorag setup --format json`
+   and read the `hister` object: `installed`, `running`, and `embedding`. Every
+   problem has a `remediation` string with the exact next command.
+2. **Install Hister** when `installed` is false: `brew install hister` (other
+   platforms: the release binaries at https://github.com/asciimoo/hister/releases).
+3. **Run it as a login item** when `running` is false: `brew services start hister`.
+   Hister must stay running: its browser extension only captures while it is up,
+   and a page visited while it is down is not re-sent. AutoRAG never starts it.
+4. **Install the browser extension** (Chrome Web Store "Hister", or Firefox
+   Add-ons). It must be done in the user's browser; tell the user if you cannot.
+5. **Share the embedding model** (below).
+6. **Before importing past history, install a skip list** in Hister's `rules.json`
+   (login, mail, documents, social and console domains, private hosts, token-bearing
+   URLs). `hister import browser history` re-downloads each URL *without* the
+   browser's cookies. See docs/datasource-skills.md, "Browser history via Hister".
+7. Add `"browser": { "type": "hister", "connector": { "semantic": true } }` under
+   `datasources`, then `autorag refresh` (only probes Hister) and
+   `autorag lite retrieve "<a question>" --scope '/browser/**' --json`.
+
+**Shared embedding model.** Hister keeps its own vector store and embeds with
+whatever its `config.yml` `semantic_search` block says. Pointing it at a different
+model does not fail: vectors simply stop lining up and results get worse. So Hister
+must use the same model as AutoRAG, which is the `minsync serve` daemon (model
+`Qwen/Qwen3-Embedding-0.6B`, 1024 dimensions, with Qwen3's query instruction).
+
+```bash
+autorag setup --format json            # reports hister.embedding.differs: the keys that disagree
+autorag setup --sync-hister            # writes them, restarts Hister, re-embeds pages if needed
+```
+
+- `--sync-hister` rewrites only `enable`, `embedding_endpoint`, `embedding_model`,
+  `dimensions`, `query_prefix`, `document_prefix` in Hister's config (a
+  `config.yml.bak.<time>` is kept next to it), then runs `brew services restart hister`,
+  waits for it to answer, and runs `hister reindex` when a change makes stored
+  vectors stale. Without the flag setup never writes Hister's config.
+- The daemon port is chosen when `minsync serve` starts. After restarting the daemon,
+  run `autorag setup --sync-hister` again. It is idempotent: when nothing differs
+  it changes nothing and restarts nothing.
+- It refuses, and says why, when the shared backend is a remote endpoint (browsing
+  text would leave the machine) or MinSync's in-process engine (Hister cannot reach
+  it; run `minsync serve --idle-timeout 0`, docs/minsync-setup.md).
+- If `restartRequired` or `reindexRequired` is true in the report, the step was
+  not done (no Homebrew service, or it failed): run `brew services restart hister`
+  and `hister reindex` yourself. Check with `hister doctor`: the line
+  `server.index.embeddings` must be `ok`.
+- A changed `query_prefix` alone needs a restart but no reindex. Never hand-edit the
+  `query_prefix` value across two lines: Hister's YAML parser folds the newline into
+  a space and the model's instruction is silently corrupted. `--sync-hister` writes
+  it correctly.
+
 ## Verify and build indexes
 
 Configuration alone is not a successful setup:
@@ -415,7 +487,8 @@ autorag refresh --json
 autorag search "summarize the collection" --top-k 3 --json --debug
 ```
 
-- `setup` probes runtime health, model profile, and datasource readiness.
+- `setup` probes the embedding backend (`embedding.backend`), MinSync, and
+  datasource readiness.
 - `status` is model-free and path-opaque.
 - `health` resolves the single model, checks credential presence, and normally
   performs one live completion probe.

package/dist/agent/peer-target-tool.d.ts

@@ -1,19 +0,0 @@
-import type { AgentTool } from "@earendil-works/pi-agent-core";
-import { Type } from "typebox";
-export declare const RECOMMEND_PEER_TARGETS_TOOL_NAME = "recommend_peer_targets";
-declare const recommendPeerTargetsSchema: Type.TObject<{
-    query: Type.TString;
-}>;
-export interface PeerTargetRecommendation {
-    readonly alias: string;
-    readonly matchedTerms: readonly string[];
-    readonly displayName?: string;
-    readonly description?: string;
-}
-export interface RecommendPeerTargetsDetails {
-    readonly method: "recommend_peer_targets";
-    readonly resultCount: number;
-    readonly matches: readonly PeerTargetRecommendation[];
-}
-export declare function createRecommendPeerTargetsTool(workspacePath: string): AgentTool<typeof recommendPeerTargetsSchema, RecommendPeerTargetsDetails>;
-export {};

package/dist/agent/query-peer-tool.d.ts

@@ -1,51 +0,0 @@
-import type { AgentTool } from "@earendil-works/pi-agent-core";
-import { Type } from "typebox";
-import { type SimplexQueryState } from "../p2p/simplex-query-store.js";
-import type { SimplexTransport } from "../p2p/simplex-transport.js";
-import type { PeerQueryResponse } from "../p2p/wire.js";
-import { EvidenceLedger } from "./evidence-ledger.js";
-export declare const QUERY_PEER_AGENT_TOOL_NAME = "query_peer_agent";
-declare const queryPeerAgentSchema: Type.TObject<{
-    alias: Type.TString;
-    query: Type.TString;
-    topK: Type.TOptional<Type.TInteger>;
-    scope: Type.TOptional<Type.TString>;
-}>;
-export interface QueryPeerAgentResult {
-    readonly number: number;
-    readonly title: string;
-    readonly summary: string;
-    readonly source: string;
-    readonly excerpt: string;
-}
-export interface QueryPeerAgentDetails {
-    readonly method: "query_peer_agent";
-    readonly ok: boolean;
-    readonly alias: string;
-    readonly contactId?: number;
-    readonly pendingId?: string;
-    readonly expiresAt?: string;
-    readonly status?: "ok" | "rejected" | "pending";
-    readonly answer?: string;
-    readonly results?: readonly QueryPeerAgentResult[];
-    readonly diagnostics?: PeerQueryResponse["diagnostics"];
-    readonly fileCount?: number;
-    readonly message?: string;
-}
-export interface QueryPeerAgentToolOptions {
-    readonly workspacePath: string;
-    readonly openTransport: () => Promise<SimplexTransport>;
-    readonly fastTimeoutMs?: number;
-    readonly autoStart?: boolean;
-    readonly sessionId?: () => string | undefined;
-    readonly onResponse?: (state: SimplexQueryState, response: PeerQueryResponse) => void;
-    readonly onExpired?: (state: SimplexQueryState) => void;
-    /** Run evidence ledger; each peer result is registered so the final emit can cite it by id. */
-    readonly ledger?: EvidenceLedger;
-}
-export interface QueryPeerAgentTool extends AgentTool<typeof queryPeerAgentSchema, QueryPeerAgentDetails> {
-    readonly ready: Promise<void>;
-    close(): Promise<void>;
-}
-export declare function createQueryPeerAgentTool(options: QueryPeerAgentToolOptions): QueryPeerAgentTool;
-export {};

package/dist/cli/commands/gateway.d.ts

@@ -1,7 +0,0 @@
-import { type RuntimeStatus } from "../../embedding-runtime/index.js";
-import type { CommandContext } from "./types.js";
-export interface GatewayCommandDeps {
-    runtimeStatus?: () => Promise<RuntimeStatus>;
-    stopRuntime?: () => Promise<void>;
-}
-export declare function runGateway(ctx: CommandContext, deps?: GatewayCommandDeps): Promise<number>;

package/dist/cli/commands/p2p-policy.d.ts

@@ -1,2 +0,0 @@
-import type { CommandContext } from "./types.js";
-export declare function runP2pPolicy(ctx: CommandContext): number;

package/dist/cli/commands/p2p.d.ts

@@ -1,7 +0,0 @@
-import type { CommandContext } from "./types.js";
-/** Resolve the workspace used for P2P identity/policy, matching `autorag serve`. */
-export declare function resolveP2pWorkspace(ctx: CommandContext): string;
-/**
- * `autorag p2p` — SimpleX peer trust management.
- */
-export declare function runP2p(ctx: CommandContext): Promise<number>;

package/dist/cli/commands/serve.d.ts

@@ -1,19 +0,0 @@
-import type { InjectionClassifierModel } from "../../p2p/injection-classifier.js";
-import { type SimplexPeerServer, startSimplexPeerServer, syncSimplexPeers } from "../../p2p/simplex-server.js";
-import { type SimplexTransport, type StartSimplexOptions } from "../../p2p/simplex-transport.js";
-import { type CliConfig, type ResolveAgentModelOptions, type ResolvedAgentModel } from "../config.js";
-import type { CommandContext } from "./types.js";
-export interface ServeCommandDeps {
-    readonly startSimplexPeerServer?: typeof startSimplexPeerServer;
-    readonly startSimplexChat?: (options: StartSimplexOptions) => Promise<SimplexTransport>;
-    readonly syncSimplexPeers?: typeof syncSimplexPeers;
-    readonly waitUntilStopped?: (server: SimplexPeerServer) => Promise<void>;
-    readonly modelResolver?: (config: CliConfig, options?: ResolveAgentModelOptions) => ResolvedAgentModel | Promise<ResolvedAgentModel>;
-}
-export declare function createInjectionClassifierModel(resolved: ResolvedAgentModel): InjectionClassifierModel;
-/**
- * `autorag serve` — start the P2P peer query server over SimpleX Chat.
- * Peers reach this installation through SimpleX; all security gates run
- * unchanged.
- */
-export declare function runServe(ctx: CommandContext, deps?: ServeCommandDeps): Promise<number>;

package/dist/datasource/skills/obsidian/client.d.ts

@@ -1,15 +0,0 @@
-import type { QmdEmbedResult, QmdEnsureResult, QmdOptions, QmdSearchMode, QmdSearchOptions, QmdSearchResult, QmdUpdateResult } from "./types.js";
-/**
- * Thin external `qmd` CLI wrapper for Obsidian vault indexing/search.
- * Every method spawns the `qmd` binary, never throws for expected failures,
- * and isolates config/cache under the AutoRAG workspace.
- */
-export declare class QmdClient {
-    private readonly options;
-    constructor(options?: QmdOptions);
-    ensureCollection(signal?: AbortSignal): Promise<QmdEnsureResult>;
-    update(signal?: AbortSignal): Promise<QmdUpdateResult>;
-    embed(signal?: AbortSignal): Promise<QmdEmbedResult>;
-    search(mode: QmdSearchMode, query: string, options?: QmdSearchOptions): Promise<QmdSearchResult>;
-    private run;
-}

package/dist/datasource/skills/obsidian/types.d.ts

@@ -1,92 +0,0 @@
-import type { RetrievalOptions } from "../../../retrieval/types.js";
-/**
- * Search modes accepted by the external `qmd` CLI.
- *
- * - `search`  — BM25 / FTS5 lexical search
- * - `vsearch` — dense vector semantic search
- */
-export type QmdSearchMode = "search" | "vsearch";
-/**
- * Configuration for {@link QmdClient}. The client spawns the `qmd` binary as a
- * child process and isolates config/cache under the AutoRAG workspace.
- */
-export interface QmdOptions {
-    /** Explicit path to the `qmd` binary. Defaults to a bare `qmd` PATH lookup. */
-    readonly binaryPath?: string;
-    /** Spawn timeout in milliseconds. Default 60_000 (embed can be slow). */
-    readonly timeoutMs?: number;
-    /** Max stdout/stderr bytes retained. Default 1_048_576 (1 MiB). */
-    readonly maxBufferBytes?: number;
-    /** Absolute path to the Obsidian vault. Required for ensure/update. */
-    readonly vaultPath?: string;
-    /** AutoRAG workspace root used to isolate qmd config/cache. */
-    readonly workspaceRoot?: string;
-    /** Instance id used for source paths and collection naming. Default `default`. */
-    readonly instanceId?: string;
-    /** Explicit collection name; defaults to a sanitized instance id. */
-    readonly collectionName?: string;
-    /** Environment overrides merged on top of a restricted process env. */
-    readonly env?: Readonly<Record<string, string | undefined>>;
-    /** Explicit operator-owned qmd config directory, never rewritten. */
-    readonly configPath?: string;
-}
-export declare const DEFAULT_QMD_BINARY = "qmd";
-export declare const DEFAULT_QMD_TIMEOUT_MS = 60000;
-export declare const DEFAULT_QMD_MAX_BUFFER_BYTES = 1048576;
-export type QmdFailureReason = "binary-missing" | "not-configured" | "nonzero-exit" | "spawn-error" | "timeout" | "aborted" | "stdout-too-large" | "stderr-too-large" | "invalid-json" | "invalid-shape";
-export interface QmdFailure {
-    readonly ok: false;
-    readonly reason: QmdFailureReason;
-    readonly stdout: string;
-    readonly stderr: string;
-    readonly code: number | null;
-}
-export interface QmdOk<T> {
-    readonly ok: true;
-    readonly data: T;
-    readonly stdout: string;
-    readonly stderr: string;
-    readonly code: number;
-}
-export interface QmdUpdateInfo {
-    readonly indexed: number;
-    readonly updated: number;
-    readonly unchanged: number;
-    readonly removed: number;
-    readonly needsEmbedding?: boolean;
-    readonly metadata?: Readonly<Record<string, unknown>>;
-}
-export interface QmdEmbedInfo {
-    readonly embedded: boolean;
-    readonly metadata?: Readonly<Record<string, unknown>>;
-}
-export interface QmdEnsureInfo {
-    readonly collectionName: string;
-    readonly vaultPath: string;
-    readonly configDir: string;
-}
-/** A scored search hit from `qmd search` / `qmd vsearch`. */
-export interface QmdSearchHit {
-    readonly chunkId: string;
-    readonly score: number;
-    readonly content: string;
-    readonly title?: string;
-    readonly file?: string;
-    readonly docid?: string;
-    readonly metadata?: Readonly<Record<string, unknown>>;
-}
-export interface QmdSearchOk {
-    readonly ok: true;
-    readonly hits: readonly QmdSearchHit[];
-    readonly stdout: string;
-    readonly stderr: string;
-    readonly code: number;
-    readonly data: {
-        readonly hits: readonly QmdSearchHit[];
-    };
-}
-export type QmdEnsureResult = QmdOk<QmdEnsureInfo> | QmdFailure;
-export type QmdUpdateResult = QmdOk<QmdUpdateInfo> | QmdFailure;
-export type QmdEmbedResult = QmdOk<QmdEmbedInfo> | QmdFailure;
-export type QmdSearchResult = QmdSearchOk | QmdFailure;
-export type QmdSearchOptions = RetrievalOptions;

package/dist/embedding-runtime/cache.d.ts

@@ -1,30 +0,0 @@
-export { CacheError } from "./types.js";
-export interface CacheAsset {
-    readonly id: string;
-    readonly url: string;
-    readonly filename: string;
-    readonly sha256: string;
-    readonly kind?: "model" | "runtime";
-    readonly archiveMembers?: readonly string[];
-}
-export declare function cacheDirectory(kind: "models" | "runtime", cacheRoot?: string): string;
-export interface CacheOptions {
-    readonly cacheRoot?: string;
-    readonly offline?: boolean;
-    readonly fetch?: typeof globalThis.fetch;
-    /** Total budget (ms) for a single asset download. Defaults to 10 minutes so large models can transfer. */
-    readonly downloadTimeoutMs?: number;
-}
-export declare function verifyCacheEntry(path: string, expectedSha256: string): Promise<string>;
-/**
- * Marker a distributor writes inside `<archive>.extracted/` when it ships the
- * runtime already extracted instead of as the pinned archive. Its content is the
- * archive's SHA-256, so a tree provisioned for another runtime build is never
- * taken for this one. A signed and notarized macOS app needs this: every Mach-O
- * it carries must be re-signed with the app's Developer ID, so the binaries can
- * no longer be packed back into an archive that matches the pinned digest.
- */
-export declare const PROVISIONED_RUNTIME_MARKER = ".autorag-archive-sha256";
-export declare function hasVerifiedAsset(asset: CacheAsset, cacheRoot?: string): Promise<boolean>;
-export declare function downloadAsset(asset: CacheAsset, options?: CacheOptions): Promise<string>;
-export declare function importAsset(sourcePath: string, asset: CacheAsset, options?: CacheOptions): Promise<string>;

package/dist/embedding-runtime/gateway-embedder.d.ts

@@ -1,41 +0,0 @@
-import type { ProfileId } from "./types.js";
-export interface EmbeddingIdentity {
-    readonly provider: string;
-    readonly model: string;
-    readonly dimension: number;
-}
-/** Embedder contract: identity is async because the gateway resolves it lazily. */
-export interface Embedder {
-    identity(): Promise<EmbeddingIdentity>;
-    embed(texts: readonly string[]): Promise<readonly (readonly number[])[]>;
-}
-export interface GatewayEmbedderOptions {
-    /** Runtime resolver; defaults to the shared embedding-runtime ensureRuntime. */
-    readonly runtime?: {
-        ensureRuntime(input?: {
-            profileId?: ProfileId;
-            cachedOnly?: boolean;
-        }): Promise<{
-            baseUrl: string;
-            identity: {
-                provider: string;
-                model: string;
-                dimension: number;
-            };
-        }>;
-    };
-    readonly profileId?: ProfileId;
-    readonly fetchImpl?: typeof fetch;
-}
-/**
- * Texts per `/v1/embeddings` call. The llama.cpp upstream rejects very large
- * batches (HTTP 413 on a full-archive sync), so embedding fans out in
- * bounded sequential batches while preserving input order.
- */
-export declare const EMBED_BATCH_SIZE = 32;
-/**
- * Production embedder backed by the loopback autorag-gateway. The runtime is
- * ensured lazily with `cachedOnly` so embedding never triggers a model
- * download; an unprimed cache surfaces as a caller-side degrade.
- */
-export declare function createGatewayEmbedder(options?: GatewayEmbedderOptions): Embedder;

package/dist/embedding-runtime/gateway.d.ts

@@ -1,27 +0,0 @@
-import type { RuntimeProfile } from "./types.js";
-export type GatewayErrorCode = "non-loopback-host" | "non-loopback-upstream" | "bad-request" | "batch-limit" | "upstream" | "upstream-violation" | "timeout";
-export declare class GatewayError extends Error {
-    readonly code: GatewayErrorCode;
-    readonly status: number;
-    constructor(code: GatewayErrorCode, message: string, status?: number, options?: {
-        readonly cause?: unknown;
-    });
-}
-export interface StartEmbeddingGatewayOptions {
-    readonly profile: RuntimeProfile;
-    readonly host?: string;
-    readonly port?: number;
-    readonly upstreamUrl: string;
-    readonly fetch?: typeof fetch;
-    readonly maxInputs?: number;
-    readonly maxCharacters?: number;
-    readonly timeoutMs?: number;
-}
-export interface EmbeddingGateway {
-    readonly host: string;
-    readonly port: number;
-    readonly url: string;
-    close(): Promise<void>;
-}
-export declare function isLoopbackHost(host: string): boolean;
-export declare function startEmbeddingGateway(options: StartEmbeddingGatewayOptions): Promise<EmbeddingGateway>;

package/dist/embedding-runtime/index.d.ts

@@ -1,120 +0,0 @@
-import { type CacheOptions } from "./cache.js";
-import { startEmbeddingGateway } from "./gateway.js";
-import { type SupervisorStatus } from "./supervisor.js";
-import type { BackendKind, HealthStatus, ProfileId, RuntimeProfile } from "./types.js";
-export interface EmbeddingRuntimeCache {
-    downloadAsset(asset: {
-        id: string;
-        url: string;
-        filename: string;
-        sha256: string;
-        kind?: "model" | "runtime";
-        archiveMembers?: readonly string[];
-    }, options?: CacheOptions): Promise<string>;
-    importAsset(sourcePath: string, asset: {
-        id: string;
-        url: string;
-        filename: string;
-        sha256: string;
-        kind?: "model" | "runtime";
-        archiveMembers?: readonly string[];
-    }, options?: CacheOptions): Promise<string>;
-    verifyCacheEntry(path: string, expectedSha256: string): Promise<string>;
-    hasVerifiedAsset?(asset: {
-        id: string;
-        url: string;
-        filename: string;
-        sha256: string;
-        kind?: "model" | "runtime";
-        archiveMembers?: readonly string[];
-    }, cacheRoot?: string): Promise<boolean>;
-}
-export interface EmbeddingRuntimeSupervisor {
-    ensureRunning(): Promise<SupervisorStatus>;
-    shutdown(): Promise<void>;
-    status(): SupervisorStatus;
-}
-export interface EmbeddingRuntimeGateway {
-    readonly url: string;
-    close(): Promise<void>;
-}
-export interface EmbeddingRuntimeOptions {
-    cacheRoot?: string;
-    backend?: BackendKind;
-    offline?: boolean;
-    platform?: "darwin-arm64-metal" | "win-x64-cpu" | "win-x64-vulkan";
-    fetch?: typeof fetch;
-    supervisor?: EmbeddingRuntimeSupervisor;
-    cache?: EmbeddingRuntimeCache;
-    gatewayFactory?: (options: Parameters<typeof startEmbeddingGateway>[0]) => Promise<EmbeddingRuntimeGateway>;
-}
-export interface EnsureRuntimeOptions extends EmbeddingRuntimeOptions {
-    profileId?: ProfileId;
-    /** Refuse downloads; used by MinSync refresh/query paths. */
-    cachedOnly?: boolean;
-}
-export interface RuntimeIdentity {
-    readonly profileId: ProfileId;
-    readonly provider: string;
-    readonly model: string;
-    readonly dimension: number;
-    readonly runtimeBuild: string;
-    readonly modelRevision: string;
-    readonly artifactSha256: string;
-}
-export interface EnsuredRuntime {
-    readonly baseUrl: string;
-    readonly profile: RuntimeProfile;
-    readonly identity: RuntimeIdentity;
-    readonly supervisor: SupervisorStatus;
-}
-export interface RuntimeStatus {
-    readonly state: SupervisorStatus["state"];
-    readonly backend: BackendKind;
-    readonly model: string;
-    readonly pid?: number;
-    readonly port?: number;
-    readonly uptimeMs?: number;
-    readonly lastError?: string;
-    readonly profileId?: ProfileId;
-    readonly baseUrl?: string;
-    readonly health: HealthStatus;
-}
-export declare function createEmbeddingRuntime(options?: EmbeddingRuntimeOptions): {
-    resolveRuntimeProfile: (profileId?: ProfileId) => RuntimeProfile;
-    ensureRuntime: (input?: EnsureRuntimeOptions) => Promise<EnsuredRuntime>;
-    releaseRuntimeHandles: () => Promise<void>;
-    stopRuntime: () => Promise<void>;
-    runtimeStatus: () => Promise<RuntimeStatus>;
-    prefetchModel: (profileId?: ProfileId) => Promise<{
-        profileId: "qwen3-embedding-0.6b";
-        path: string;
-    }>;
-    importModel: (profileId: ProfileId | undefined, sourcePath: string) => Promise<{
-        profileId: "qwen3-embedding-0.6b";
-        path: string;
-    }>;
-    verifyModel: (profileId?: ProfileId) => Promise<{
-        profileId: "qwen3-embedding-0.6b";
-        path: string;
-        hash: string;
-    }>;
-};
-export declare const resolveRuntimeProfile: (profileId?: ProfileId) => RuntimeProfile;
-export declare const ensureRuntime: (input?: EnsureRuntimeOptions) => Promise<EnsuredRuntime>;
-export declare const releaseRuntimeHandles: () => Promise<void>;
-export declare const stopRuntime: () => Promise<void>;
-export declare const runtimeStatus: () => Promise<RuntimeStatus>;
-export declare const prefetchModel: (profileId?: ProfileId) => Promise<{
-    profileId: "qwen3-embedding-0.6b";
-    path: string;
-}>;
-export declare const importModel: (profileId: ProfileId | undefined, sourcePath: string) => Promise<{
-    profileId: "qwen3-embedding-0.6b";
-    path: string;
-}>;
-export declare const verifyModel: (profileId?: ProfileId) => Promise<{
-    profileId: "qwen3-embedding-0.6b";
-    path: string;
-    hash: string;
-}>;

package/dist/embedding-runtime/manifest.d.ts

@@ -1,20 +0,0 @@
-import type { BackendKind, PlatformId, ProfileId, RuntimeAsset, RuntimeProfile } from "./types.js";
-export declare function resolveProfile(profileId: ProfileId): RuntimeProfile;
-export declare function platformRuntimeAsset(platform: PlatformId): RuntimeAsset;
-export declare function selectPlatformAsset(platform: PlatformId, backend: BackendKind, vulkanAvailable?: boolean): RuntimeAsset;
-export declare const MODEL_ASSETS: {
-    readonly qwen3: {
-        readonly url: "https://huggingface.co/Qwen/Qwen3-Embedding-0.6B-GGUF/resolve/370f27d7550e0def9b39c1f16d3fbaa13aa67728/Qwen3-Embedding-0.6B-Q8_0.gguf";
-        readonly revision: "370f27d7550e0def9b39c1f16d3fbaa13aa67728";
-        readonly sha256: "06507c7b42688469c4e7298b0a1e16deff06caf291cf0a5b278c308249c3e439";
-        readonly license: "Apache-2.0";
-        readonly noticeReference: "Qwen3 model card and Apache-2.0 license";
-        readonly intendedUse: "Local loopback embedding for AutoRAG MinSync and native-datasource semantic retrieval. Not a generative model.";
-        readonly unsuitableUse: "Chat or generation, remote embedding of corpus text, shipping weights in the npm package, any GGUF other than the pinned SHA-256.";
-        readonly upstreamCardUrl: "https://huggingface.co/Qwen/Qwen3-Embedding-0.6B";
-        readonly modelCard: "docs/model-cards/qwen3-embedding-0.6b.md";
-    };
-};
-export declare const MODEL_ASSET_URLS: {
-    readonly qwen3: "https://huggingface.co/Qwen/Qwen3-Embedding-0.6B-GGUF/resolve/370f27d7550e0def9b39c1f16d3fbaa13aa67728/Qwen3-Embedding-0.6B-Q8_0.gguf";
-};