{"generated_at": "2026-10-26T19:47:13.214176+00:00", "entries": [{"name": "0bull", "repo": "https://github.com/0bull/agent-plugin", "sha": "240890a71715375bc877b238491e414517e9d902", "description": "Control real iPhones rented from 0bull (screenshots, on-screen text, taps, typing, device commands, recorded macros, an on-phone agent) and publish videos to TikTok and YouTube Shorts from them, through the hosted 0bull Streamable HTTP MCP server plus three workflow skills. Portable Agent Plugins v1 package (mcp.json + skills/), no plugin code. Requires a 0bull account with a rented phone and browser OAuth consent (add the server with auth: oauth, then hermes mcp login 0bull); no API key or client secret is stored in the plugin. Posting sends the user's video and caption to TikTok or YouTube through the phone's own apps. Disclosure \u2014 hosted-only connector; screenshots, on-screen text, anything typed on the phone (including passwords or codes you hand the agent) and your videos and captions go to 0bull.net; posts publish publicly to TikTok or YouTube; the server also exposes rental tools that spend money, which the skills tell the model never to call but Hermes does not block.", "maintainer": "0bull", "tier": "community", "category": "tools", "docs_url": "https://github.com/0bull/agent-plugin#readme", "version": "1.0.0", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "aave-mcp", "title": "Aave", "repo": "https://github.com/aave/skills", "sha": "7132f6e952f9100f16577c33774a1e63948f7fab", "subdir": "plugins/mcp", "version": "1.0.0", "description": "Aave V3 and V4 through Aave's official MCP server (Streamable HTTP at https://mcp.aave.com, no sign-in, no keys in config): live markets and rates on every chain Aave runs on, wallet positions and health factor, rewards, DAO governance, swap quotes, and unsigned transactions for the user's own wallet to sign. Bundles five workflow skills: safe transactions, deleverage, yield analysis, account activity and transaction confirmation. Agent Plugins v1 package (plugin.json + mcp.json + skills/), configuration only. Disclosure: hosted-only connector with no local code; every tool call goes to mcp.aave.com, which records tool usage for analytics and keeps a wallet address only as a keyed hash. The server never holds keys. Every tool is annotated read-only except submit_signed_order and cancel_order, which relay an order or cancellation the user already signed; the package marks the server untrusted, so Hermes asks before either runs and refuses them in unattended runs. The tool set is served by the remote server at runtime and is not covered by the pinned sha.", "maintainer": "Aave Labs", "tier": "community", "category": "tools", "docs_url": "https://github.com/aave/skills/tree/main/plugins/mcp", "requires_hermes": ">=0.21.6", "readme": true, "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "abstract-provider", "title": "Abstract", "repo": "https://github.com/abstractpbc/hermes-abstract-provider", "sha": "e322e42578586af1c376b7e836dfce437308682a", "description": "Abstract model provider (GLM-5.3, DeepSeek-V4-Flash, Kimi-K3) over the OpenAI Responses API with a bundled model catalog. Disclosure \u2014 registers one model provider at import and nothing else (no tools, hooks, commands or threads); requests go to https://api.abstract.inc/v1 through Hermes' own client with ABSTRACT_API_KEY, which Hermes reads; the plugin itself opens no connections, reads and writes no files and starts no processes.", "maintainer": "abstractpbc", "tier": "community", "category": "models", "requires_hermes": ">=0.22.2", "version": "1.0.0", "docs_url": "https://github.com/abstractpbc/hermes-abstract-provider#readme", "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": ["ABSTRACT_API_KEY"]}}, {"name": "accent-picker", "repo": "https://github.com/NousResearch/hermes-official-plugins", "sha": "abec9e822a23089943dd1d476a104bf746f19f86", "subdir": "accent-picker", "title": "Accent Picker", "description": "OKLCH accent colour picker for the Hermes Desktop status bar: drag it and the whole theme retints live, with palette commands to reset to the theme default or copy the current colour. A theme-authoring tool; the colour is scratch state and is never persisted. Desktop-only (desktop/plugin.js inside the plugin SDK surface); no agent tools, hooks, middleware or env vars.", "maintainer": "NousResearch", "tier": "official", "category": "desktop", "version": "1.0.0", "docs_url": "https://github.com/NousResearch/hermes-official-plugins/tree/main/accent-picker", "image": "https://raw.githubusercontent.com/NousResearch/hermes-official-plugins/abec9e822a23089943dd1d476a104bf746f19f86/assets/accent-picker.jpg", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "account-pool", "repo": "https://github.com/roqianjas/hermes-account-pool", "sha": "7a4fc9fd976ed68de9e6f70ba326ef2053741e69", "subdir": "plugins/account-pool", "version": "4.0.2", "title": "Account Pool", "description": "Run several accounts of one provider in Hermes' credential pool: `hermes account ` adds accounts with email verification, shows live usage per account, keeps the least-used account first (fill_first + a 5-minute balance timer over every provider with accounts), parks accounts for hours or on a weekly schedule, shows which bot is on which account and its tokens at API prices, alerts on dead logins, a full pool or a runaway bot, and can start an idle account's 5-hour window with one tiny request (wake). Adapters for anthropic and openai-codex; generic commands for any pool provider. Optional Accounts page in Hermes Desktop. Linux and Windows. Disclosure \u2014 logins stay in Hermes: park holds an entry out of rotation in Hermes' pool (mark_exhausted_and_rotate, reason \"parked\") and unpark is reset_status; the plugin's park list holds labels, ids and times only; expired tokens are refreshed by Hermes (try_refresh_matching); sign-in, in the terminal and on the Desktop page, runs `hermes auth add`; other profiles' accounts are read with read_credential_pool. It uses each account's access token as Hermes stores it for live usage and the email check (api.anthropic.com/api/oauth/usage and /profile, chatgpt.com/backend-api/wham/usage, with the client headers of Hermes' usage fetcher) and for wake (a command, a page button, or the timer after the opt-in `wake auto-on`): one one-word request per account through Hermes' Anthropic client or the Codex Responses endpoint, and the account's model list. Telegram alerts through `hermes send -t telegram` to the bot's home channel or a chat id kept in account-pool.json; no telemetry. Its session tracker, a post_api_request hook, runs in every profile and writes labels, ids and token counts only. `setup` writes plugins.enabled of every profile, links the plugin into every profile, removes the retired account-pool-tracker (`hermes plugins remove`) and, with --timer, adds a systemd user timer (Linux) or a Task Scheduler task for the user (Windows, no admin rights). Runs systemctl --user and notify-send (Linux), schtasks, powershell.exe and taskkill (Windows), hermes config set, hermes auth add/remove, hermes send, hermes plugins remove; reads every profile's config.yaml and state.db (read-only), the gateway config (whether the Telegram bot has a home channel) and /proc on Linux. The command registers in the default profile only and provides no tools. Live usage, the email check and wake send each account's own access token from Hermes' pool straight to its vendor with Claude Code (anthropic) and Codex CLI (openai-codex) client headers. The session tracker reads the running agent's current pool account (label and id only) by scanning Python thread frames for Hermes' private _credential_pool. On the first run after 3.x it moves parked logins from account-pool-parked.json back into Hermes' pool and deletes that file.", "maintainer": "roqianjas", "tier": "community", "category": "tools", "requires_hermes": ">=0.21.6", "docs_url": "https://github.com/roqianjas/hermes-account-pool/blob/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/README.md", "platforms": ["linux", "windows"], "screenshots": ["https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/accounts.png", "https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/report.png", "https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/sessions.png", "https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/alerts.png", "https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/add-account.png", "https://raw.githubusercontent.com/roqianjas/hermes-account-pool/7a4fc9fd976ed68de9e6f70ba326ef2053741e69/docs/screenshots/add-account-codex.png"], "known_issues": ["After installing, run `hermes account setup` (add --timer for the 5-minute balance timer): it lists the plugin in every profile for the Desktop page and links it into every profile for the session tracker. Coming from 3.x it also removes account-pool-tracker.", "Hermes can let a parked account back into rotation early (`hermes auth reset`, or a Codex quota it sees come back); the next balance run parks it again, and doctor names it until then. A park without --until is held for 30 days, and a balance run renews it once less than 15 days are left, so without the timer it ends after 30 days.", "Windows: tested with a stand-in Task Scheduler and, up to 3.13, with Windows' own Python, cmd.exe and wscript.exe under Wine; not yet on a Windows machine. On Windows the timer runs only while the user is signed in.", "Built for plugins.isolation: in_process (the default). Under host, Hermes skips plugin CLI commands, so `hermes account` and the timer's balance are unavailable.", "Usage comes from undocumented provider endpoints (api.anthropic.com/api/oauth/usage, chatgpt.com/backend-api/wham/usage) that can change or rate-limit without notice; bots are unaffected when they do.", "The Desktop page's sign-in relays `hermes auth add` and has run in sandbox tests (Hermes' real sign-in code, provider calls replaced), not yet with a new account on a real install. The Codex browser sign-in needs the browser on the same computer; when port 1455 is taken, Hermes signs in with a device code instead."], "capabilities": {"provides_tools": [], "provides_hooks": ["post_api_request"], "provides_middleware": [], "requires_env": []}}, {"name": "account-switcher", "repo": "https://github.com/SmelterLabs/hermes-account-switcher", "sha": "f31ad5e715fa32a3c0350d8fcf8ee787b224d105", "subdir": "", "description": "Windows desktop plugin that switches which Codex login and which Claude subscription account every local Hermes profile uses, with one idle-only, verified restart of the gateway (and, on Hermes before 0.22, of Hermes Desktop). Refuses while any profile is busy, restores the previous selection if a step fails, and writes a receipt that says whether the switch took effect. Disclosure \u2014 Windows only. A switch rewrites, through Hermes's own commands, the Codex credential order in every profile's auth.json and the CLAUDE_SUBSCRIPTION_DIRECTSDK_CONFIG_DIR line in every profile's .env, so every profile that has the plugin enabled spends the same selected Codex login and the same Claude login (one Hermes-owned claude-auth folder; no token is copied between profiles). It reads Codex tokens from each profile's auth.json. On Hermes 0.22 and later it stops and restarts the gateway once Hermes's own idle proof and the gateway's drain report idle, and Hermes Desktop stays open and reconnects; on earlier Hermes it also reads the Desktop backends' dashboard session token from their process environment and closes and reopens Hermes Desktop (work started in the seconds after the idle check is cut off). It sends the target Codex token to OpenAI's usage endpoint and Claude login tokens to Anthropic's OAuth profile endpoint (also when the setup screen opens); if the target Claude login has expired it first runs the claude CLI once so the CLI renews its own token, which rotates that login's refresh token. No telemetry and no automatic account rotation.", "maintainer": "SmelterLabs", "tier": "community", "category": "desktop", "requires_hermes": ">=0.21.6", "docs_url": "https://github.com/SmelterLabs/hermes-account-switcher#readme", "version": "1.3.2", "platforms": ["windows"], "readme": true, "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "acedatacloud-media", "title": "Ace Data Cloud Media", "repo": "https://github.com/AceDataCloud/AceDataCloudMediaHermes", "sha": "8eaebfb86152d754061375de881548bfa096dd2d", "description": "Suno music, Seedance video, HappyHorse (Happy Horse) video, GPT Image images and Nano Banana images by Ace Data Cloud; 17 selectable service toolsets and a Nano Banana image_generate backend. Each generate call spends Ace Data Cloud credits on your ACEDATACLOUD_API_KEY with no Hermes approval prompt; the model can choose costly options (pro models, 4K, up to 4 images), and prompts and reference URLs go to api.acedata.cloud. Google Search and face analysis also spend Credits. Disclosure \u2014 Each generate call spends Ace Data Cloud credits on your ACEDATACLOUD_API_KEY with no Hermes approval prompt, and the model can pick costly options (pro models, 4K, up to 4 outputs); Google Search and face analysis are billed too. Prompts, reference URLs and face images go to api.acedata.cloud. The tools are on wherever you enable their toolsets, including gateways.", "maintainer": "AceDataCloud", "tier": "community", "category": "tools", "version": "2.0.2", "requires_hermes": ">=0.21.6", "image": "https://raw.githubusercontent.com/AceDataCloud/AceDataCloudMediaHermes/8eaebfb86152d754061375de881548bfa096dd2d/assets/catalog.png", "readme": true, "capabilities": {"provides_tools": ["acedatacloud_nano_banana_generate", "acedatacloud_nano_banana_task", "acedatacloud_veo_generate", "acedatacloud_veo_task", "acedatacloud_suno_generate", "acedatacloud_suno_task", "acedatacloud_fish_generate", "acedatacloud_fish_task", "acedatacloud_gpt_image_generate", "acedatacloud_gpt_image_task", "acedatacloud_seedance_generate", "acedatacloud_seedance_task", "acedatacloud_seedream_generate", "acedatacloud_seedream_task", "acedatacloud_happyhorse_generate", "acedatacloud_happyhorse_task", "acedatacloud_qwen_image_generate", "acedatacloud_qwen_image_task", "acedatacloud_google_search_search", "acedatacloud_grok_generate", "acedatacloud_grok_task", "acedatacloud_face_analyze", "acedatacloud_midjourney_generate", "acedatacloud_midjourney_task", "acedatacloud_flux_generate", "acedatacloud_flux_task", "acedatacloud_minimax_generate", "acedatacloud_minimax_task", "acedatacloud_wan_generate", "acedatacloud_wan_task", "acedatacloud_kling_generate", "acedatacloud_kling_task"], "provides_hooks": [], "provides_middleware": [], "requires_env": ["ACEDATACLOUD_API_KEY"]}, "docs_url": "https://github.com/AceDataCloud/AceDataCloudMediaHermes/blob/8eaebfb86152d754061375de881548bfa096dd2d/README.md"}, {"name": "acedatacloud", "title": "Ace Data Cloud", "repo": "https://github.com/AceDataCloud/AceDataCloudModelProviderHermes", "sha": "568763ee21fe081562bb9bcd91e7323f8f81d6ff", "image": "https://raw.githubusercontent.com/AceDataCloud/AceDataCloudModelProviderHermes/568763ee21fe081562bb9bcd91e7323f8f81d6ff/assets/catalog.png", "description": "Ace Data Cloud GPT models with streaming and tool calling.", "maintainer": "AceDataCloud", "tier": "community", "category": "models", "requires_hermes": ">=0.21.5", "version": "1.0.2", "docs_url": "https://github.com/AceDataCloud/AceDataCloudModelProviderHermes/blob/568763ee21fe081562bb9bcd91e7323f8f81d6ff/README.md", "known_issues": ["The model picker currently offers gpt-4.1-mini and gpt-4.1 only."], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": ["ACEDATACLOUD_API_KEY"]}}, {"name": "actvox", "title": "ActVox", "repo": "https://github.com/ActVox/actvox-plugins", "sha": "76e394d381c3cf56632d7fd6c91d442c68b181f2", "subdir": "hermes", "version": "0.1.1", "description": "Make Hermes a team member in an ActVox workspace: on a schedule it works the tasks people assign it and answers the comments that @mention it, and submits each task for a person's review. Skills actvox:actvox and actvox:actvox-inbox, a first-turn inbox line (pre_llm_call), /actvox inbox|status, and hermes actvox setup|install-gate. A cron pre-run gate wakes the agent only when the inbox has work, so an empty tick makes no model call. Disclosure \u2014 ActVox is reached only through the user's own actvox MCP server (ctx.call_mcp, needs plugins.entries.actvox.mcp_allowlist) and the user's own actvox CLI (the gate runs `actvox inbox --all --json`); on the first turn of every non-cron session the pre_llm_call hook calls the server's check_inbox (up to 5 s) and adds one line saying how many tasks and mentions wait; the plugin holds and reads no credential and sends no telemetry. `hermes actvox install-gate` writes one file, $HERMES_HOME/scripts/actvox-inbox-gate.py, only on that explicit command and never over a file it did not write; gate state lives in $HERMES_HOME/plugin-data/actvox/. No background work of its own: the cron job runs only after the user creates it, and from then on anyone who can assign Hermes a task or @mention it in your ActVox workspaces decides what an unattended agent run on your host works on, with that job's toolsets (dangerous commands follow approvals.cron_mode, deny by default). Unattended writes need the default trust: full on the actvox server.", "maintainer": "ActVox", "tier": "community", "category": "automation", "docs_url": "https://github.com/ActVox/actvox-plugins/tree/main/hermes", "requires_hermes": ">=0.22.5", "readme": true, "platforms": [], "known_issues": ["The inbox gate needs Node.js and the ActVox CLI (npm i -g @actvox/cli) on the gateway host.", "With plugins.isolation: host, the `hermes actvox` terminal commands are not available."], "capabilities": {"provides_tools": [], "provides_hooks": ["pre_llm_call"], "provides_middleware": [], "requires_env": []}}, {"name": "adaption", "repo": "https://github.com/adaptionlabs/adaption-hermes-plugin", "sha": "958ad2e9077107be1919a5365541caa4e0acdcc8", "description": "Dataset import, adaptation, augmentation, translation, synthetic generation, and AutoScientist fine-tuning through the hosted Adaption MCP. Portable Agent Plugins v1 package with three workflow skills; the MCP server is added separately with OAuth (or an API key for headless use). Disclosure \u2014 skill-only, no code; the MCP tools call api.prod.adaptionlabs.ai, and launches spend Adaption credits (the skills request a cost estimate first, Hermes does not enforce it). Hermes does not prompt before those tools unless the server is configured with trust: untrusted.", "maintainer": "adaptionlabs", "tier": "community", "category": "tools", "requires_hermes": ">=0.21.5", "docs_url": "https://github.com/adaptionlabs/adaption-hermes-plugin#readme", "version": "0.1.0", "readme": true, "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "adaptive-inference-routing", "repo": "https://github.com/midhatn/adaptive-inference", "sha": "d6f4e201ba28602189872746922e527c58ed0cbd", "subdir": "plugins/hermes", "title": "Adaptive Inference Routing", "description": "Opt-in background scheduling for host-known delegated sessions, supported cron execution and explicit worker profiles using an existing Adaptive Inference gateway. Interactive and explicitly classified requests retain their routing; model, tools, prompts, output limits and credentials are unchanged. Disclosure: adds only X-Adaptive-Class on an exactly configured chat-completions endpoint. No network calls, file access, processes, telemetry or credential handling in plugin code. Uses a bounded in-memory session index. Cron detection requires the newer public execution-identity capability; auxiliary routing is an explicit provider configuration, not an auxiliary middleware claim.", "maintainer": "midhatn", "tier": "community", "category": "models", "requires_hermes": ">=0.21.5", "docs_url": "https://github.com/midhatn/adaptive-inference/blob/d6f4e201ba28602189872746922e527c58ed0cbd/plugins/hermes/README.md", "version": "0.1.0", "capabilities": {"provides_tools": [], "provides_hooks": ["subagent_start", "subagent_stop", "post_compaction"], "provides_middleware": ["llm_request"], "requires_env": []}}, {"name": "adkit", "title": "AdKit", "repo": "https://github.com/adkit/ads-mcp", "sha": "97f9a776fd317329f544d10ce22792eb4b115c62", "subdir": "hermes", "version": "1.0.0", "description": "Manage ads on Google Ads, Meta Ads (Facebook, Instagram), TikTok, LinkedIn, Microsoft, Reddit and X Ads through the hosted AdKit MCP: launch campaigns, ad sets and ads as drafts, pull performance reports with breakdowns, browse competitor ads in the Meta, Google and LinkedIn ad libraries, and generate ad creatives. Built for people who run several ad accounts. Agent Plugins v1 package (plugin.json + mcp.json), configuration only; OAuth uses pinned mcp-remote@0.1.49 over npx stdio, so Node.js 20.18.1+ and npm are required. Disclosure \u2014 creates and edits real ad campaigns that spend your ad budgets (creates are drafts until you approve publishing); the package marks the adkit server untrusted, so Hermes asks before every tool the server does not annotate as read-only and refuses them in cron and headless runs. Every tool call goes to mcp.adkit.so; sign-in runs npx mcp-remote@0.1.49 from the npm registry, which stores its OAuth tokens in ~/.mcp-auth. The tool set is served by the remote server at runtime and is not covered by the pinned sha.", "maintainer": "AdKit", "tier": "community", "category": "tools", "requires_hermes": ">=0.21.6", "docs_url": "https://github.com/adkit/ads-mcp/tree/main/hermes#readme", "image": "https://raw.githubusercontent.com/adkit/ads-mcp/97f9a776fd317329f544d10ce22792eb4b115c62/assets/logo.png", "readme": true, "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "adspirer", "repo": "https://github.com/Adspirer/adspirer-hermes-plugin", "sha": "4688c38295a7b86f8119d7a0ecc09cd0b83a2730", "description": "Create, analyze, and optimize campaigns across Google Ads, Meta Ads, TikTok Ads, LinkedIn Ads, Amazon Ads, ChatGPT Ads, and Microsoft Ads through the hosted Adspirer MCP. Portable Agent Plugins v1 package with fourteen paid-media workflow skills and approval safeguards; OAuth uses pinned mcp-remote@0.1.49 over npx stdio, so Node.js 20.18.1+ and npm are required. Disclosure \u2014 creates and edits real ad campaigns that spend your ad budgets; confirm-before-spend is enforced by skill instructions, not by Hermes.", "maintainer": "Adspirer", "tier": "community", "category": "tools", "requires_hermes": ">=0.20", "docs_url": "https://github.com/Adspirer/adspirer-hermes-plugin#readme", "version": "1.0.0", "image": "https://raw.githubusercontent.com/Adspirer/adspirer-hermes-plugin/4688c38295a7b86f8119d7a0ecc09cd0b83a2730/assets/icon.png", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "advanced-delegation", "repo": "https://github.com/NousResearch/hermes-official-plugins", "sha": "e404b6816b20237a82f6fbe2749b5a576cf75dba", "subdir": "advanced-delegation", "title": "Advanced Delegation", "description": "Two extra delegation modes for `delegate_task`, next to the built-in default. Model routes (`delegation.backend: model_map`): the agent picks a named model route per subagent from a list you define (e.g. fast, deep, local). Profile routes (`delegation.backend: profile_routes`): the agent runs a subagent as one of your Hermes profiles, with that profile's own model, credentials, skills and SOUL.md. Only route names and their one-line descriptions reach the model. Adds `hermes delegation config` (pick the mode and edit its routes; `--show` prints the setup) and the `/delegation` slash command (save a mode for the next session, or `--now` to apply it to the current chat). Needs Hermes 0.22.5 or newer, the first release with pluggable delegation providers; Hermes 0.22.0 to 0.22.4 refuses to install it, and on a core older than 0.22.0 it installs, logs one warning and does nothing. Disclosure \u2014 no Python dependencies and no network or model calls of its own; subagents run through core's `delegate_task` on the routes you configured. Writes only `delegation.backend` and `delegation.backends.*` (model_map / profile_routes) in the active profile's config.yaml, through core's config writer. A model route's optional `api_key` is stored in config.yaml in plain text (`provider:` keeps keys in .env / auth.json). Listing a profile under profile routes lets this profile's subagents use that profile's credentials, SOUL.md and terminal policy.", "maintainer": "NousResearch", "tier": "official", "category": "tools", "requires_hermes": ">=0.22.5", "version": "1.0.0", "image": "https://raw.githubusercontent.com/NousResearch/hermes-official-plugins/e404b6816b20237a82f6fbe2749b5a576cf75dba/assets/advanced-delegation.jpg", "docs_url": "https://github.com/NousResearch/hermes-official-plugins/tree/main/advanced-delegation", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "advisor", "repo": "https://github.com/ljubomirj/hermes-advisor-plugin", "sha": "37e55975611d38212cf082029ece504f207df656", "version": "1.2.0", "title": "Advisor", "description": "A persistent second model reviews every agent turn. After each turn the plugin hands the transcript to the model you nominate as advisor and delivers its findings as advice in the same session: [NIT], [CONCERN] or [BLOCKER] notes, held back until a later review raises the same point again, so the primary agent is not interrupted by a one-off remark. It reviews, it does not act: the advisor never edits files, calls tools or changes session state, and only reads the turn transcript. This is the backend counterpart of MoA \u2014 one extra model, not a committee: the primary agent does the work and the advisor checks it. Port of pi-omplike-advisor (Petr Baudis / pasky), itself rooted in oh-my-pi (Can Boluk). Runs inside the plugin host with no dependencies beyond Hermes itself. Disclosure \u2014 once the plugin is enabled, every completed turn costs one extra completion through Hermes's ctx.llm to the advisor model (the primary agent's provider unless you allow a provider/model override under plugins.entries.advisor.llm), carrying the user message, the final reply, every tool call and tool result in the session so far (each clipped to 500 characters) and the contents of a WATCHDOG.md in Hermes's working directory if one exists; deliverable advice is queued into the CLI session as a user-role message, which starts or interrupts an agent turn; state is kept under $HERMES_HOME/advisor/ and reviews run on one background thread; no other network, files or credentials.", "maintainer": "ljubomirj", "tier": "community", "category": "automation", "requires_hermes": ">=0.21", "docs_url": "https://github.com/ljubomirj/hermes-advisor-plugin/blob/37e55975611d38212cf082029ece504f207df656/README.md", "readme": true, "platforms": ["linux", "macos", "windows"], "capabilities": {"provides_tools": [], "provides_hooks": ["post_llm_call", "on_session_finalize"], "provides_middleware": [], "requires_env": []}, "known_issues": ["Every reviewed turn costs one extra completion on the advisor model. Reviews run on a single worker in the background after the hook returns (the hook itself never blocks turn finalization), so a long advisor call delays advice, not the agent.", "Advice is injected into the live conversation with ctx.inject_message, which the interactive CLI provides. On gateway surfaces (Telegram, Slack, ...) the review still runs and its notes are held, but there is no in-session place to inject them, so delivery is CLI-shaped.", "Concerns are only delivered once a later review repeats them (hold-and-reconfirm), so a genuine one-off concern can stay silent by design. /advisor test injects a synthetic note to see the delivery path.", "The cross-provider trust gate is conservative: point the advisor at a provider different from the primary agent's, or accept the default inherit-primary behaviour.", "The interactive provider/model picker (/advisor model with no argument) needs PluginContext.request_model_selection, and the status-bar badge needs ctx.register_status_segment. Where a build lacks those newer seams the plugin still loads and the text commands keep working (/advisor provider , /advisor model ).", "State lives in $HERMES_HOME/advisor/state.json, per profile."]}, {"name": "adwhispr", "repo": "https://github.com/adwhispr/adwhispr-hermes-plugin", "sha": "80fd0e8fd93fe6207e6e9dd288811be9fe695bc2", "description": "Research competitor ads, clone the winners, and run campaigns through the hosted AdWhispr MCP. Finds brands verified to be advertising in a niche right now, their longest-running Meta and TikTok ads, Instagram content and Google keyword data; clones a proven ad into a finished image or AI video for your brand (free script approval before any credit is charged); and launches paused campaigns on Meta, Google, TikTok, X and ChatGPT Ads. Portable Agent Plugins v1 package with twenty-four workflow skills; OAuth uses pinned mcp-remote@0.1.49 over npx stdio, so Node.js 20.18.1+ and npm are required. Disclosure \u2014 creates and edits real ad campaigns that spend your ad budgets (always created paused, two-step confirm server-side) and renders videos that use paid credits; the AdWhispr server is marked untrusted, so Hermes asks before every write or spend tool and refuses them in cron and headless runs. Sends research queries and your brand assets to adwhispr.com; sign-in is OAuth through npx mcp-remote@0.1.49.", "maintainer": "AdWhispr", "tier": "community", "category": "tools", "requires_hermes": ">=0.21.6", "docs_url": "https://github.com/adwhispr/adwhispr-hermes-plugin#readme", "version": "1.0.1", "image": "https://raw.githubusercontent.com/adwhispr/adwhispr-hermes-plugin/80fd0e8fd93fe6207e6e9dd288811be9fe695bc2/assets/icon.png", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "afterforge", "repo": "https://github.com/ussyverse/afterforge", "sha": "067a739e5ab64097140c5200bde2cc5d52662ea5", "description": "Local-first regression workbench turning agent failures and corrections into reviewed, evidence-backed checks.", "maintainer": "ussyverse", "tier": "community", "category": "tools", "docs_url": "https://github.com/ussyverse/afterforge/blob/067a739e5ab64097140c5200bde2cc5d52662ea5/README.md", "platforms": ["linux"], "capabilities": {"provides_tools": ["fixlab_status", "fixlab_scan", "fixlab_list_cases", "fixlab_inspect_case", "fixlab_review_case", "fixlab_build_regression", "fixlab_verify_regression", "fixlab_report"], "provides_hooks": ["post_tool_call", "on_session_end", "pre_verify"], "provides_middleware": [], "requires_env": []}}, {"name": "agent-analytics", "repo": "https://github.com/Agent-Analytics/agent-analytics-hermes-plugin", "sha": "63f11d3669330e40dad977b7d32f9b69e69a9fb8", "description": "Dashboard-only read plugin for Agent Analytics inside Hermes.", "maintainer": "Agent-Analytics", "tier": "community", "category": "desktop", "docs_url": "https://github.com/Agent-Analytics/agent-analytics-hermes-plugin#readme", "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "agent-batch", "repo": "https://github.com/raaaas/hermes-agent-batch", "sha": "064afc5dba8869df703ba055aa2760730c0d30dc", "subdir": "plugin", "description": "Parallel AI task orchestrator. Paste a task list into the desktop panel; Hermes phases it into dependency-ordered groups through the agent_batch_plan / agent_batch_phase tools (or you group it yourself in one click), then each phase dispatches one independent coding agent per task to a GitHub Actions workflow in a repo you name \u2014 own branch, own PR \u2014 and the panel tracks the runs and the PRs. GITHUB_TOKEN is the only required credential; AGENT_BATCH_REPO, AGENT_BATCH_WORKFLOW, AGENT_BATCH_MODEL and AGENT_BATCH_STATE are optional (the last is a directory override for the plan file, not a secret) and each can be set per plan in the panel instead. Disclosure \u2014 Hermes can read and rewrite the stored phase plan through the agent_batch_plan / agent_batch_phase tools, but dispatch is always a confirmed click in the panel. Each dispatch fires a GitHub workflow_dispatch with your GITHUB_TOKEN (Actions + pull-requests write) on the repo you name. The workflow you copy into that repo installs third-party coding-agent CLIs (cline by default; the others are unpinned npm packages) and runs them unattended, with auto-approve / skip-permissions, repo write access and your provider secrets. They push real branches and open real PRs, and your task text and repository code go to that runner's model vendor. LLM cost belongs to the runner: cline spends Cline Credits or a BYOK key, opencode free models need its zen relay, kilo needs an active Kilo Go subscription (keyless CI runs fail), and claude-code / codex / dsh need their own keys.", "maintainer": "raaaas", "tier": "community", "category": "automation", "docs_url": "https://github.com/raaaas/hermes-agent-batch", "capabilities": {"provides_tools": ["agent_batch_plan", "agent_batch_phase"], "provides_hooks": [], "provides_middleware": [], "requires_env": ["GITHUB_TOKEN"]}, "requires_hermes": ">=0.21.3", "version": "1.0.0"}, {"name": "agent-hint", "repo": "https://github.com/dachaimahogany/hermes-agent-hint", "sha": "5601d2353529352fb89c5ad85a1a9d3a68150cbb", "description": "Give the inference engine behind each request a standard, machine-readable view of the agent session that produced it. Injects an X-Agent-Hint header (base64url JSON: session identity and call index, lifecycle state, soft priority, a cache-pinning hint, optional measured statistics) on main LLM requests, and can emit session lifecycle events out of band as local ndjson or as POSTs to a URL you configure. Request bodies are never modified, and nothing is added or sent until the plugin is enabled. Lets a scheduler pin a session to the instance already holding its KV cache, or rank a user-blocked session above background work, without teaching the engine anything about the client. Disclosure \u2014 when enabled, every main LLM request carries an X-Agent-Hint header (base64url JSON with the session id, call index, lifecycle state, priority and cache hints) to whatever provider is configured, not only hint-aware engines; with status_sink: http the same session events and usage/latency statistics are POSTed to the status_url you set, and ndjson writes them under $HERMES_HOME.", "maintainer": "dachaimahogany", "tier": "community", "category": "general", "docs_url": "https://github.com/dachaimahogany/hermes-agent-hint#readme", "version": "0.5.1", "requires_hermes": ">=0.21.6", "platforms": [], "capabilities": {"provides_tools": [], "provides_hooks": ["on_session_start", "on_session_end", "on_session_finalize", "pre_tool_call", "post_tool_call", "pre_approval_request", "post_approval_response", "subagent_start", "subagent_stop", "post_api_request"], "provides_middleware": ["llm_request"], "requires_env": []}}, {"name": "agent-hold-em", "repo": "https://github.com/tony-simons-aiowa/agent-hold-em", "sha": "dcd3b8d265778a70e6e671467cf2e6ec3cf21f42", "subdir": "package/agent-hold-em", "version": "1.0.0", "description": "No-Limit Texas Hold 'Em for Hermes Desktop: play with three isolated Hermes profile agents using play chips only. Agent decisions use configured models and may consume provider tokens; this is a local game, not gambling software. Disclosure \u2014 each opponent is a zero-tool Hermes agent run in-process under the local Hermes profile you select for it, using that profile's configured provider credentials and spending its tokens; reads every local profile's name, model and avatar for the picker; stores the table (including hole cards) in plugin-data/agent-hold-em/table.json.", "maintainer": "tony-simons-aiowa", "tier": "community", "category": "desktop", "screenshots": ["https://raw.githubusercontent.com/tony-simons-aiowa/agent-hold-em/dcd3b8d265778a70e6e671467cf2e6ec3cf21f42/assets/catalog-setup.png"], "capabilities": {"provides_tools": [], "provides_hooks": [], "provides_middleware": [], "requires_env": []}}, {"name": "agent-importers", "repo": "https://github.com/NousResearch/hermes-official-plugins", "sha": "b5932f1b03ccd0f7738cb4975318df990408f3a3", "subdir": "agent-importers", "description": "`hermes import-agent`: import an OpenClaw (~/.openclaw, or legacy ~/.clawdbot / ~/.moltbot), Claude Code (~/.claude), Codex CLI (~/.codex) or Cursor (~/.cursor) setup into Hermes in one command; always previews first and asks before applying. Claude Code / Codex / Cursor: instructions, Cursor rules and memories become memory entries, Bash permission rules become command allow/deny lists, MCP servers land in config.yaml, skills are copied, and `--sync` re-imports sources that changed. OpenClaw (core's former `hermes claw migrate`; the plugin keeps that spelling, `hermes claw cleanup` and the openclaw-migration skill): persona, memories, skills, model and messaging settings, MCP servers and agent behaviour. Formerly built into Hermes core; `hermes setup` offers to install it when it finds one of these agents, and profiles with a registered sync source are migrated automatically. Disclosure \u2014 no Python dependencies, no network calls and no model calls. Claude Code / Codex / Cursor: reads those agents' local config trees (never their credential files; secret-looking MCP env vars and headers are stripped and listed) and writes only into the active profile (memories/MEMORY.md with a backup, config.yaml, skills/ and import-sync.json). OpenClaw: reads the OpenClaw dir (openclaw.json, its .env, workspace files such as SOUL.md / MEMORY.md / USER.md, skills); API keys and bot tokens come over only with --migrate-secrets (allowlisted names, shown in the preview). It writes into the active profile (config.yaml, .env, SOUL.md, memories/, skills/openclaw-imports/, a migration/openclaw/