> ## Documentation Index
> Fetch the complete documentation index at: https://icrl.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# StorageAdapter

> Pluggable storage interface for trajectories and embeddings

## Interface

`StorageAdapter` defines trajectory, metadata, and embedding persistence plus vector similarity search.

## Required Methods

**Init & Trajectories:**

```ts theme={null}
init(): Promise<void>
saveTrajectory(trajectory: Trajectory): Promise<void>
getTrajectory(id: string): Promise<Trajectory | null>
getAllTrajectories(): Promise<Trajectory[]>
deleteTrajectory(id: string): Promise<boolean>
```

**Curation Metadata:**

```ts theme={null}
saveCurationMetadata(meta: CurationMetadata): Promise<void>
getCurationMetadata(trajectoryId: string): Promise<CurationMetadata | null>
getAllCurationMetadata(): Promise<CurationMetadata[]>
deleteCurationMetadata(trajectoryId: string): Promise<boolean>
```

**Embeddings:**

```ts theme={null}
saveEmbedding(embedding: StoredEmbedding): Promise<void>
saveEmbeddings(embeddings: StoredEmbedding[]): Promise<void>
getEmbedding(id: string): Promise<StoredEmbedding | null>
getEmbeddingsByType(type: "trajectory" | "step"): Promise<StoredEmbedding[]>
deleteEmbeddingsForTrajectory(trajectoryId: string): Promise<void>
searchByEmbedding(query: number[], type: "trajectory" | "step", k: number): Promise<EmbeddingSearchResult[]>
```

**Step Examples:**

```ts theme={null}
getStepExamplesForTrajectory(trajectoryId: string): Promise<StepExample[]>
getAllStepExamples(): Promise<StepExample[]>
```

## Built-in Implementation

* `FileSystemAdapter(path: string, options?: FileSystemAdapterOptions)` — stores trajectories as JSON on disk with in-memory embedding index

`FileSystemAdapterOptions`:

* `normalizeEmbeddings` (default: true)