GitHub profilePlugins by yuzushi-dev
3 plugins by yuzushi-dev in the catalog, most-starred first.
yuzushi-devhermes-seoNative Hermes port of Claude SEO 2.4.2: 45 SEO workflow and specialist skills, a credential-scoped hermes_seo_run tool, an isolated Python runtime, and advisory JSON-LD feedback. Derived from Claude SEO 2.4.2 (MIT; FLOW prompt materials CC BY 4.0). Disclosure — derived from Claude SEO 2.4.2 (MIT; FLOW prompt materials CC BY 4.0); the hermes_seo_run tool starts local subprocesses in the Hermes host process, not a Docker terminal sandbox, and model-supplied script arguments may name host paths, so it can read host files (the Gemini image-edit utility uploads the named file to Google, the IndexNow utility reads a URL-list file); `hermes seo setup` downloads Python packages from PyPI within version ranges into a private environment and optionally Chromium through Playwright; audits fetch user-selected websites; separately configured providers (Moz, Bing Webmaster, Keywords Everywhere, Google APIs, Matomo, DataForSEO, IndexNow, Google Gemini) receive the data sent to them and may incur charges; skills ask before paid or mutating calls but the tool does not enforce approvals; Gemini images are saved to ~/Documents/nanobanana_generated, and other runtime state stays in Hermes plugin data storage; provider keys are read from the active Hermes profile and passed only to the script that needs them; Google OAuth workflows are disabled; no telemetry, bundled credentials, background process or self-updater.Communitytools126yuzushi-devhermes-handoffProfile-bound semantic session handoffs for CLI and Desktop with independent destination and optional summary-writer models. Derived from yuzushi-dev/session-handoff. Disclosure — reads bounded persisted session evidence from the originating profile's state.db and project/Git metadata; runs local Git and foreground Hermes subprocesses without a shell; stores private handoff/journal sidecars under that profile; optional writer and new chat turns use configured model providers and may incur inference costs. No telemetry or plugin-owned credential store. Four internal tool-policy imports remain version-coupled; no runtime overrides or self-updater.Communityautomationyuzushi-devrelic-gumiA companion that gets to know you and supports self-reflection, built on published psychometric instruments (TIPI for the Big Five, ECR-RS for attachment) as a self-report baseline plus consented check-ins, with an output safety critic and Telegram or Desktop onboarding. Not a diagnostic tool; based on the paper arXiv:2611.08504. Disclosure — state lives under $RELIC_HOME (default ~/.relic); the installer creates a dedicated relic-gumi profile, edits that profile's config.yaml (enables the plugin, sets cron.wrap_response false, copies the base model provider/model/base_url but never keys), creates an empty .env there that the plugin never reads or writes, moves a relic plugin copy found in the base Hermes home to a backup and removes it from the base plugins.enabled list, and creates Hermes cron jobs in that profile that stay silent until the user consents during onboarding; with the "learn from what you write" consent it reads the profile's own Hermes state.db user messages to update a self-report profile; credentials (TELEGRAM_BOT_TOKEN, TELEGRAM_ALLOWED_USERS, TELEGRAM_HOME_CHANNEL, GEMINI_API_KEY, OLLAMA_API_KEYS) are optional plugin secrets read only through Hermes' secret scope; check-ins are delivered by detached subprocesses (job and scoped credentials over a private stdin pipe, minimal environment with proxy/CA settings) straight to the Telegram Bot API; network is the configured Hermes provider, Telegram, local Ollama classifiers, Google Gemini/Lyria media only with RELIC_ENABLE_GOOGLE_MEDIA=1 and a key, Ollama Cloud (ollama.com) only if explicitly configured, and local ComfyUI or ACE-Step servers only if the user enables them; no telemetry, no self-updater.Communitygeneral
