No user-submitted related servers found.
Scored across 143 tools
Many tools have overlapping purposes within the same domain: four variants of sending messages (openclaw_send, openclaw_agent, openclaw_chat_send, openclaw_sessions_send), two abort tools (openclaw_chat_abort, openclaw_sessions_abort), and multiple approval-policy tools (openclaw_exec_approval_* vs openclaw_exec_approvals_*) that are easy to confuse. The descriptions are detailed, but the sheer number of near-miss CRUD operations makes misselection likely.
Naming is a mix of domain_action (openclaw_cron_list), domain_subdomain_action (openclaw_node_pair_list), camelCase method names (openclaw_tts_setProvider), and plural/singular inconsistencies (openclaw_exec_approval_list vs openclaw_exec_approvals_get). While the openclaw_ prefix provides a uniform namespace, the structure after it is inconsistent enough to hurt predictability.
143 tools is an extreme count for a single MCP server, far beyond even a heavy 25+ threshold. While the gateway has a large surface area, exposing every JSON-RPC method plus convenience wrappers and escape hatches results in a sprawling toolset that is overwhelming for an agent to navigate.
The toolset is remarkably comprehensive across gateway domains: sessions, agents, config, cron, skills, nodes, approvals, TTS, and more. Missing operations like skill uninstall, a pending-node list, or a session message history (only tail/preview) are minor gaps; the coverage is strong enough that most workflows have no dead ends.