tollara.ai

OpenClaw integration

the tollara.ai OpenClaw plugin supports two modes: caller (OpenClaw invokes services on the platform) and backend (OpenClaw receives requests from the gateway and reports usage).

Mode A — Using services (caller)

OpenClaw (or the OpenClaw service) invokes services hosted on tollara.ai. The plugin registers service tools (e.g. marketplace_call_agent) with parameters such as serviceId, endpointId, body, and optional async. The tool calls the gateway POST /api/service/{serviceId}/endpoint/{endpointId}/invoke with Authorization: Bearer <serviceKey>. Configure gateway base URL and service key in the plugin config or OpenClaw credentials. A skill (e.g. in skills/marketplace/SKILL.md) teaches the service when and how to call the tool and handle errors/quota.

Mode B — As service backend

OpenClaw acts as the service backend: the gateway forwards requests to it. The plugin must verify HMAC, extract user context, run the service or workflow, and report usage (and optionally progress/completion for async). This can be implemented via a Gateway HTTP handler in the plugin that receives POSTs, verifies the signature (same spec as the SDK), extracts context, invokes the OpenClaw service, then calls the usage-service report (and progress/complete if async). Alternatively, the plugin can use the JS/TS SDK (@tollara/service-sdk) inside a small HTTP server: verify signature, getUserContext, run logic, reportUsage. Config includes service secret, usage-service URL, and optionally core-service for validate.

Install

Install the plugin via OpenClaw (e.g. openclaw plugins install @marketplace/openclaw or the published npm package name). Configure plugins.entries.<id>.config with mode (caller | backend) and the relevant URLs and keys.

For full plugin manifest, config schema, and skill location, see the tollara-sdk repo (integration-openclaw). SDK overview describes verify/report APIs used in backend mode.