API Playground

The API Playground is a workspace-scoped test harness for Slab5 API calls. It is meant for verification before wiring a real client or agent into Slab5.

Purpose

Use the playground to test the API surface exposed to workspace API keys:

• Select a category and endpoint.
• Inspect the required scopes for the selected operation.
• Fill path variables such as :asset_id.
• Send requests through the active environment's API route.
• Reuse IDs returned from create calls in follow-up requests.
• Test asset storage upload, completion, and download flows.
• Test Vector Store/RAG collections, AgentGrid workflow discovery, AgentGrid memory records, and Slab5 Guide session creation when those endpoints are available in the selected environment.

The playground uses the active workspace context from the control plane. It still requires a workspace API key so the request exercises the same authentication, scope, module, audit, usage, and billing checks as an external API client.

Authentication

Create a workspace API key from API Keys, choose the smallest scope set needed for the test, and paste the key into the playground's workspace API key field.

The key is used only for playground requests in the current browser session. Do not paste production keys into screenshots, issue comments, docs examples, or support tickets.

If you want app-level attribution in Usage or Audit Log, create the key from the Registered App detail view or associate the key with the app before testing.

Endpoint categories

The playground groups routes by product area. Current categories include:

Category	Examples	Typical scopes
Core modules	CRM, Support, Tasks, Projects, Products, Calendar, CMS, Assets, Activity Log	Module-specific :read or :write scopes
Data and insights	Data Agents, BI queries, analytics governance, Object Analytics	analytics:read, analytics:write, or Data Agent scopes
AgentGrid	Runs, tool calls, retrieval hits, costs, workflow discovery, memory records, memory embeddings	AgentGrid and related module scopes
RAG in a box	Vector collection list/create/archive, document upsert/delete, search	vector_store:read, vector_store:write
Assistants	Slab5 Guide session creation	Assistant-specific route availability
Integrations	Webhooks and external connection-related routes	Integration scopes

Endpoint availability can depend on workspace modules, scopes, plan limits, and environment configuration.

Endpoint testing

Pick a category, then select an endpoint. The playground loads the method, path, request body template, required scopes, and any path-variable fields.

For create operations, a successful response can populate the returned data.id into the matching follow-up field. For example, creating an asset can populate asset_id for the complete, get, or archive calls.

If a request fails, use the response request_id and structured error code to inspect audit, usage, logs, and troubleshooting docs.

RAG and Vector Store testing

Use the RAG in a box category to validate managed Vector Store operations before writing client code.

Recommended test order:

1. GET /v1/vector-collections?limit=10
2. POST /v1/vector-collections
3. POST /v1/vector-collections/:collection_key/documents
4. POST /v1/vector-collections/:collection_key/search
5. DELETE /v1/vector-collections/:collection_key/documents
6. DELETE /v1/vector-collections/:collection_key

After the test, open Usage → Intelligence usage and confirm the collection-level Vector Store events were recorded. If the key is app-associated, open the app detail page and confirm the collection and recent usage are attributed to the app.

AgentGrid testing

Use the AgentGrid category to test workflow and memory endpoints that are exposed through the app-plane API.

Useful checks:

• create or list runs
• create retrieval hits and cost events for traceability
• discover similar workflows with POST /v1/agent-grid/workflow-discovery
• reindex workflow discovery with POST /v1/agent-grid/workflow-discovery/reindex
• create and update memory records
• create and update memory search-index bindings

Workflow discovery and memory features use Slab5 managed retrieval where configured. Review Usage → Intelligence usage to see AgentGrid workflow discovery and AgentGrid memory usage events.

Asset storage upload

For Assets, the playground includes a Storage upload panel:

1. Select Assets.
2. Create an asset upload intent with POST /v1/assets.
3. Confirm the response includes upload_url, upload_method, upload_headers, and upload_expires_at.
4. Select a local file in Storage upload, or use the text body for a tiny test object.
5. Click PUT to upload bytes directly to the returned signed URL.
6. Call POST /v1/assets/:asset_id/complete to mark the asset available.
7. Call GET /v1/assets/:asset_id and open the returned download_url.

When a file is selected, the playground uses the file bytes for the upload, fills the content type from the file when available, and uses the file size and MIME type when preparing the complete call.