The agent that grows with you
Python aiai-agentai-agentsanthropicchatgptclaudeclaude-codecodexhermeshermes-agentllmnous-researchopenai
chore(catalog): review disclosure for onebot-platform
Tekniuma4f027b75720
diff --git a/plugin-catalog/onebot-platform.yaml b/plugin-catalog/onebot-platform.yamlindex e648c470e1c..8891a4282d7 100644--- a/plugin-catalog/onebot-platform.yaml+++ b/plugin-catalog/onebot-platform.yaml@@ -9,26 +9,7 @@ category: platformrequires_hermes: ">=0.22.2"docs_url: https://github.com/mario841859784/hermes-plugin-onebot/blob/086dd7e5c13004e41301f5cea1f352c46b0f3c50/README.mdplatforms: []-description: >-- QQ bot via the OneBot 11 protocol: connects Hermes to your own self-hosted- NapCat, Lagrange, LLOneBot or go-cqhttp deployment over WebSocket — the- plugin hosts a reverse-WS server inside the Hermes gateway, or dials an- existing forward-WS endpoint — and adds the qq toolset (qq_send_image,- qq_send_voice, qq_send_video, qq_send_file, qq_send_forward, qq_napcat_api,- qq_group_history) for DM and group chats with admin allowlists, group- mention gating, per-member soft limits, and cron/notification delivery to a- configured home channel. Inbound images are downloaded and (with Pillow)- downscaled before the vision tool sees them; long replies are split at- sentence boundaries or rendered locally into a text image with- fonttools/Pillow. Disclosure — talks only to the user's own OneBot- deployment; the WebSocket listener is a long-running background process- that lives inside the Hermes gateway process; inbound message images are- downloaded from that deployment into the plugin's data directory for the- vision tool; text-to-image rendering runs locally (fonttools, Pillow,- system CJK fonts) with no third-party calls; credentials (the OneBot- access token, allowed-user lists, home channel) reach the plugin only- through Hermes's ONEBOT_* env/config channel and go nowhere but your own- OneBot deployment; no telemetry or usage reporting.+description: "QQ bot via the OneBot 11 protocol: connects Hermes to your own self-hosted NapCat, Lagrange, LLOneBot or go-cqhttp deployment over WebSocket — the plugin hosts a reverse-WS server inside the Hermes gateway, or dials an existing forward-WS endpoint — and adds the qq toolset (qq_send_image, qq_send_voice, qq_send_video, qq_send_file, qq_send_forward, qq_napcat_api, qq_group_history) for DM and group chats with admin allowlists, group mention gating, per-member soft limits, and cron/notification delivery to a configured home channel. Inbound images are downloaded and (with Pillow) downscaled before the vision tool sees them; long replies are split at sentence boundaries or rendered locally into a text image with fonttools/Pillow. Disclosure — talks only to the user's own OneBot deployment; the plugin's listener (default port 8643) is a long-running background server inside the Hermes gateway process that accepts the bridge's reverse WebSocket and also serves the qq_* tools' local HTTP API, refuses a non-loopback bind without an access token, and limits tool sends from a QQ session to that chat or the configured home channel unless the session is an admin user's private chat; inbound message images are downloaded from that deployment into the plugin's data directory for the vision tool; text-to-image rendering runs locally (fonttools, Pillow, system CJK fonts) with no third-party calls; credentials (the OneBot access token, allowed-user lists, home channel) reach the plugin only through Hermes's secret scope and ONEBOT_* config and go nowhere but your own OneBot deployment; no telemetry or usage reporting."capabilities:provides_tools:- qq_send_image