Three MCP servers for WorkerKit, served over Streamable HTTP.
Manage your AI worker fleet: list workers with live run state, trigger and watch runs, curate memory (rules & facts), manage schedules and instructions, start/stop workers, install kits from the public directory as new workers, and call the app tools an Action worker exposes, with no model in between.
https://mcp.workerkit.ai/workers
Authentication: connect via OAuth from your MCP client, or send a WorkerKit
manager key directly as Authorization: Bearer pe_mgr_…. An account
admin creates manager keys under Fleet access.
Discover, search and read worker reports, reconcile report changes, and read dated
source observations on a separately enabled account timeline. Use the same manager
keys and sign-in as the workers server. Reads need readKnowledge;
status updates need manageKnowledge and do not act in source apps.
Configuration and timeline activation also require an account administrator and an unrestricted credential. Enabling the timeline authorizes bounded source-excerpt retention and paid model extraction. Observations are incomplete; verify live sources before acting.
https://mcp.workerkit.ai/knowledge
Browse and search the public WorkerKit kits directory — pre-built AI workers with pre-scoped app permissions: search by job, app, category or model, inspect a kit's permissions and instruction, and get its install page.
https://mcp.workerkit.ai/directory
claude mcp add --transport http workerkit https://mcp.workerkit.ai/workers claude mcp add --transport http workerkit-knowledge https://mcp.workerkit.ai/knowledge claude mcp add --transport http workerkit-directory https://mcp.workerkit.ai/directory
© WorkerKit.ai