909 plugins from the official Hermes plugin catalog: 20 official, 9 verified and 880 from the community. 206 were added in the last week. Search, filter by what you want to do, and copy the install command.
880 plugins. Sorted by HermesWatcher's picks: usefulness, polish and trust. Official: made by Nous Research. Verified: listed with a verified maintainer. Community: built by others, so read the source first.
Brave Search
Community
Brave Search API tools for live web search inside Hermes
Brave Search API tools for live web search inside Hermes
How to install
Install it. Hermes asks for any API keys it needs (this one needs name), then asks whether to turn it on.
hermes plugins install brave-search
Said no? Turn it on later:
hermes plugins enable brave-search
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Fail-closed attribution middleware for Sabi's local adaptive routing lane (sabi-code). Adds opaque session/turn headers on loopback only; ignores all other models. Verified on Hermes 0.21.3.
How to install
Install it. Hermes asks for any API keys it needs (this one needs SABI_HERMES_BASE_URL), then asks whether to turn it on.
hermes plugins install sabi-metadata
Said no? Turn it on later:
hermes plugins enable sabi-metadata
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Checkbox staged memory writes and approve or reject the selection from a dialog.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install memory-review
Said no? Turn it on later:
hermes plugins enable memory-review
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Woof
Community
Plan, launch, watch and compare WOOF weather simulations on your own NVIDIA GPU: size runs with WOOF's own planner, launch them locally or on an ssh/WSL GPU host, get maps in the chat while the forecast runs, compare…
Plan, launch, watch and compare WOOF weather simulations on your own NVIDIA GPU: size runs with WOOF's own planner, launch them locally or on an ssh/WSL GPU host, get maps in the chat while the forecast runs, compare physics or resolution, run multi-arm research campaigns and score runs against radar and surface observations.
Good to know before installing
Runs need Linux x86_64 with an NVIDIA GPU (plan-only elsewhere); the user-run hermes woof install pip-installs recast-woof>=1.0.0,<1.1 into ~/woof-env and downloads WOOF tables and optional geography (about 1.6 GB); runs execute the WOOF CLI as detached background jobs writing to ~/woof-runs and fetch public weather data; run events are pushed into the chat as user-role messages; the optional serve/ scripts download llama.cpp and a Qwen GGUF and start an unauthenticated local model server.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install woof
Said no? Turn it on later:
hermes plugins enable woof
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Gives ChatGPT/Codex-subscription models the context window the backend actually accepts instead of the one its catalog advertises. Measured live on 2026-10-14: gpt-6.1-sol advertises 500K, caps at 872K and accepts 919,956 input tokens, while the -900k opt-in variant had silently stopped working because Hermes only raises a legacy 272K advertisement. The plugin reads the live catalog, writes one scoped model_overrides.<provider>.<model_id>.context_window entry per model (a documented step of context resolution, so no core patch), and ships a probe that bisects the API for the true ceiling because the refusal is a late SSE event, not an HTTP error. It also selects core's Codex-backed image provider so image_generate runs on the subscription with no OPENAI_API_KEY.
Good to know before installing
Uses the profile's Codex sign-in through Hermes's credentialed_provider_request (requires_auth: [openai-codex]); the plugin never reads auth.json and never holds a token, and network calls go only to chatgpt.com. Its only writes are scoped model_overrides entries and image_gen.provider (only when unset), through Hermes's atomic config writers; revert removes exactly what it wrote. No telemetry, no background processes, no shell commands, no self-updater. probe spends subscription quota (refused requests are free, accepted ones bill) and is never run automatically. Disclosure — uses the profile's Codex sign-in only through Hermes's credentialed_provider_request (requires_auth: openai-codex) and calls chatgpt.com only; its CLI and the model-callable codex_truly tool write scoped model_overrides context_window entries and image_gen.provider (only when unset) to config.yaml with no approval prompt, and `probe` sends large padding requests that spend subscription quota.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install codex-truly
Said no? Turn it on later:
hermes plugins enable codex-truly
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Nexus
Community
Self-hosted long-term memory on Qdrant: automatic per-turn prefetch and write-back, plus recall/remember/forget, guardrail rules and a knowledge graph; Hermes and any MCP-connected agent share one collection.
Self-hosted long-term memory on Qdrant: automatic per-turn prefetch and write-back, plus recall/remember/forget, guardrail rules and a knowledge graph; Hermes and any MCP-connected agent share one collection. Embeddings run fully local by default (sentence-transformers with Qwen/Qwen3-Embedding-0.6B, 1024d, about 1.2 GB from Hugging Face on first use; on Apple Silicon the model is pinned to the CPU, because PyTorch's MPS backend aborts the whole process under two overlapping encodes). A cloud embedder runs only when NEXUS_EMBEDDING_PROVIDER names it and its key is present, and a key that merely sits in the environment never selects one. Access levels fail closed: a memory, a protection rule or a graph label without a known level counts as the most restrictive. A local service on 127.0.0.1:9122 serves embeddings and the full MCP toolset; its /embed and /mcp routes demand a per-installation key (service_auth.key, mode 0600) and reject non-loopback Host/Origin, while the open /healthz answers liveness only. The plugin starts that service when Hermes cannot load the embedding stack, and it keeps running afterwards. Session-end fact extraction sends conversation text to the model endpoint from Hermes config.yaml, and the consolidation daemon to its configured model, which defaults to Hermes's own default model through local Ollama; reads ~/.hermes/.env and config.yaml; state and daily backups under ~/.nexus-memory; NEXUS_UPDATE_CHECK=1 opts into a GitHub release check.
Good to know before installing
When Hermes cannot load the embedding stack (and always on Apple Silicon unless NEXUS_EMBED_IN_HOST=1), the plugin starts `nexus-memory serve` on 127.0.0.1:9122 (embeddings, the full Nexus MCP toolset and a consolidation daemon that sends memory text to the Ollama endpoint at 127.0.0.1:11434 or NEXUS_OLLAMA_BASE under the model named by NEXUS_CONSOLIDATION_MODEL, nexus-memory.consolidation_model or Hermes's model.default, which can be an Ollama cloud model), and the service keeps running after Hermes exits; its /embed and MCP routes require a per-install key in ~/.nexus-memory/service_auth.key (0600) and a loopback Host/Origin; session-end fact extraction sends conversation text to the model endpoint in Hermes config.yaml, and the opt-in recall reranker (nexus-memory.rerank or NEXUS_RERANK=1) defaults to sending each query plus up to 20 memory snippets to that same endpoint using the provider API key it reads from config.yaml; reads ~/.hermes/.env and config.yaml; state and daily backups under ~/.nexus-memory; opt-in NEXUS_UPDATE_CHECK=1 adds a GitHub release check and an upgrade nudge to the system prompt.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install nexus
Said no? Turn it on later:
hermes plugins enable nexus
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Octen web search with query-focused highlights and page extraction.
How to install
Install it. Hermes asks for any API keys it needs (this one needs OCTEN_API_KEY), then asks whether to turn it on.
hermes plugins install web-octen
Said no? Turn it on later:
hermes plugins enable web-octen
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Opt-in bounded idle prompt warming for explicitly allowlisted OpenAI-compatible text-chat endpoints. Sends potentially billable requests; no tool execution or conversation writes. Does not support Codex/Responses or native Anthropic. Authenticated endpoints use optional HERMES_SESSION_WARMER_API_KEY.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install hermes-session-warmer
Said no? Turn it on later:
hermes plugins enable hermes-session-warmer
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Durable native memory for Hermes with LLM fact extraction. Conversations are distilled by the Hermes-configured LLM into searchable memories — facts, preferences and decisions with deduplication and importance — instead of storing turns verbatim, and are kept in Qdrant (embedded, self-hosted Server, or Qdrant Cloud) with configurable embeddings (default: local Ollama qwen3-embedding:4b at 2560 dimensions). The plugin registers a native MemoryProvider plus four scoped tools: qdrant_memory_search, qdrant_memory_add, qdrant_memory_update and qdrant_memory_delete. A SQLite WAL admission ledger makes every write idempotent and replayable across restarts with fail-closed recovery; scoped inventory, portable export and scope-aware deletion (list/export/delete-all with explicit single-agent vs all-agents semantics) ship in the same CLI, and `migrate mem0` imports existing Mem0 exports with resumable, verifiable runs.
Good to know before installing
Memory text is stored in the configured Qdrant store (embedded files under <HERMES_HOME>/qdrant-memory/qdrant by default, or the Server/Cloud URL you set), and memory text plus each turn's recall query are sent as plain text to the configured embedding endpoint (default local Ollama at http://127.0.0.1:11434; an OpenAI-compatible remote endpoint can be configured, and keyed endpoints must use HTTPS). Disclosure — after each primary-agent turn, at session end and before context compression, conversation text goes to the LLM configured in Hermes through the Hermes plugin LLM facade (a provider/model override needs the plugin's Hermes trust grant) to extract facts and to compare near-duplicate memories; that LLM may be a remote provider. Disclosure — the provider is selected with memory.provider: qdrant-memory; its settings live in <HERMES_HOME>/qdrant-memory.json (0600, secrets stripped on save). QDRANT_URL, QDRANT_API_KEY and EMBEDDING_API_KEY are read through the Hermes secret scope, but an api_key written by hand into qdrant-memory.json takes precedence, and `migrate mem0` reads the source Qdrant endpoint and any api_key from <HERMES_HOME>/mem0.json or --source-config. Disclosure
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install qdrant-memory
Said no? Turn it on later:
hermes plugins enable qdrant-memory
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Ollama Cloud web search + content extraction via Ollama's native REST API. Search returns web results; extract fetches page content. Reuses OLLAMA_API_KEY (https://ollama.com/settings/keys).
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install web-ollama-cloud
Said no? Turn it on later:
hermes plugins enable web-ollama-cloud
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Cliproxyapi Provider
Community
CLIProxyAPI as native Hermes model providers: discovers each CPA provider group (Claude, Codex, xAI, Gemini, Kiro and others) and registers one profile per group on the wire its CPA executor accepts (Anthropic…
CLIProxyAPI as native Hermes model providers: discovers each CPA provider group (Claude, Codex, xAI, Gemini, Kiro and others) and registers one profile per group on the wire its CPA executor accepts (Anthropic Messages, Responses or Chat Completions), with context windows and reasoning levels from CPA.
Good to know before installing
Sends model-list and inference requests, authenticated with the API key env var you name, to the CPA base_url you configure; with an optional management key it also makes bounded read-only GETs to CPA's management API; with an optional billing route it sends inference, and a model-list GET, to the gateway URL you configure, authenticated with the gateway key env var you name. Known limit: Hermes strips from agent subprocesses only the key env vars a registered provider declares, so the management key env var, and on a routed instance the CPA key env var, stay readable by agent shells.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install cliproxyapi-provider
Said no? Turn it on later:
hermes plugins enable cliproxyapi-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Hwiki
Community
Markdown wiki as Hermes long-term memory: plain files you and the agent edit together (Obsidian-compatible, git-versionable), local SQLite FTS5 search with optional qmd hybrid retrieval.
Markdown wiki as Hermes long-term memory: plain files you and the agent edit together (Obsidian-compatible, git-versionable), local SQLite FTS5 search with optional qmd hybrid retrieval. Inspired by the Karpathy LLM wiki.
Good to know before installing
Reads every Markdown file under the wiki folder you set (wiki_path, outside HERMES_HOME; untrusted_dirs, default raw/, excluded) and injects up to ~1,400 characters of matching excerpts into each turn; the agent appends dated notes only under <wiki>/<memory_dir> (default wiki/memory), and additions to Hermes's built-in memory are mirrored there by default (mirror_builtin_memory), so they land in a folder you may commit or sync; keeps an index at HERMES_HOME/hwiki-index.sqlite and config in HERMES_HOME/hwiki.json, and adds the wiki folder to hermes backup. Cron and subagent runs get read-only recall. No network with the default fts5 backend; the optional qmd backend runs the separately installed qmd CLI (npm @tobilu/qmd, Node ≥22), whose vector modes download local models (~0.3–1.6 GB).
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install hwiki
Said no? Turn it on later:
hermes plugins enable hwiki
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Oxylabs AI Studio web search and extraction provider for Hermes Agent. Requires OXYLABS_API_KEY (sign up at https://aistudio.oxylabs.io/api-key).
How to install
Install it. Hermes asks for any API keys it needs (this one needs name), then asks whether to turn it on.
hermes plugins install web-oxylabs
Said no? Turn it on later:
hermes plugins enable web-oxylabs
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Cursor model provider using the official proprietary Cursor SDK.
Good to know before installing
Uses Cursor's proprietary cursor-sdk (PyPI, bundled binary bridge on loopback) with the user's CURSOR_API_KEY; sends the full conversation, tool schemas/results and images (HTTPS URLs or base64) to Cursor; keeps private per-session SDK workspaces under HERMES_HOME/cache/cursor-sdk; Cursor built-in tools and MCP are requested off and Hermes keeps tool execution and approvals.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install cursor-provider
Said no? Turn it on later:
hermes plugins enable cursor-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
OptChat-style endless memory for Hermes: an append-only verbatim log plus a binary summary tree, exposed as recall through the MemoryProvider ABC. Background compactor builds the tree with a cheap auxiliary model (call_llm task optchat_compact); prefetch returns a fixed-budget view of the whole history; optchat_zoom/optchat_date tools open any line down to the verbatim message.
Good to know before installing
Every turn's user messages, agent replies, tool calls and tool results (each result capped at 30,000 characters, after Hermes's secret redaction) are appended permanently to unencrypted owner-only files under $HERMES_HOME/optchat/, with no deletion or retention limit; a background compactor sends those messages and the surrounding summary view to your auxiliary model route (task optchat_compact) for summarization; each turn injects up to view_budget_chars (default 16,000) of that history into the model context.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install optchat
Said no? Turn it on later:
hermes plugins enable optchat
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
ZeroSignal — private, pay-per-request inference through the local zs-proxy; no API key, your wallet is the credential.
How to install
Install it. Hermes asks for any API keys it needs (this one needs ZEROSIGNAL_API_KEY), then asks whether to turn it on.
hermes plugins install zerosignal
Said no? Turn it on later:
hermes plugins enable zerosignal
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Local-first persistent memory for coding agents. SQLite on your disk - decisions, preferences, error patterns, and guidelines with project and global scopes and guideline-aware search. No cloud, no API key, nothing leaves the machine.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install sugar
Said no? Turn it on later:
hermes plugins enable sugar
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Muse Spark in Hermes billed to the Muse Code monthly login (device-code login, no API key).
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install hermes-muse-code
Said no? Turn it on later:
hermes plugins enable hermes-muse-code
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Yantrikdb
Community
Persistent memory for Hermes. Canonicalizes duplicate memories, surfaces contradictions between structured or recognized single-valued claims for review, and returns a why_retrieved reason list with every recall.
Persistent memory for Hermes. Canonicalizes duplicate memories, surfaces contradictions between structured or recognized single-valued claims for review, and returns a why_retrieved reason list with every recall. Embedded by default: no server, no token, no GPU. Optional HTTP backend for HA clusters; pluggable embedders (bundled potion-2M default).
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install yantrikdb
Said no? Turn it on later:
hermes plugins enable yantrikdb
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Abstract model provider (GLM-5.3, DeepSeek-V4-Flash, Kimi-K3) over the OpenAI Responses API with a bundled model catalog.
Good to know before installing
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.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs ABSTRACT_API_KEY), then asks whether to turn it on.
hermes plugins install abstract-provider
Said no? Turn it on later:
hermes plugins enable abstract-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Memory Rewind
Community
Automatic version history for the built-in memory (MEMORY.md, USER.md), SOUL.md and skills/ (including curator-archived skills): browse, diff and restore any past version with `hermes memory-rewind`, or view it from…
Automatic version history for the built-in memory (MEMORY.md, USER.md), SOUL.md and skills/ (including curator-archived skills): browse, diff and restore any past version with `hermes memory-rewind`, or view it from any chat with the read-only `/memory-history`. Works alongside the built-in memory; not a memory provider.
Good to know before installing
Runs the local git binary (isolated from the user's git config, hooks and signing) to keep a bare repository in $HERMES_HOME/plugin-data/memory-rewind/; reads only the tracked files (never .env, auth.json, databases or skills/.hub/); rewrites tracked files only when the user runs `restore`; no network.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install memory-rewind
Said no? Turn it on later:
hermes plugins enable memory-rewind
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Model provider; prompts, tool schemas and BHARATCODE_API_KEY go to bharatcode.ai (BharatCode, India), which processes prompts through third-party inference partners, some outside India; outside paid Pro (complimentary or referral Pro included) BharatCode may retain private prompt/response logs for debugging, abuse review and product quality, and even on paid Pro a flagged account can have a time-limited sample of up to 100 prompts and responses captured for security review.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs BHARATCODE_API_KEY), then asks whether to turn it on.
hermes plugins install bharatcode-provider
Said no? Turn it on later:
hermes plugins enable bharatcode-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Self-memory and replacement models for Hermes agents. Remember your work. Train your replacement.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install icarus
Said no? Turn it on later:
hermes plugins enable icarus
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Add and manage custom OpenRouter models in Hermes without modifying the core catalog.
How to install
Install it. Hermes asks for any API keys it needs (this one needs OPENROUTER_API_KEY), then asks whether to turn it on.
hermes plugins install hopper
Said no? Turn it on later:
hermes plugins enable hopper
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Self-owned long-term memory backed by the agent-memory-es service (Elasticsearch). Three agent tools — recall, retain, reflect — over episodic/ semantic/procedural tiers with hybrid BM25+kNN recall and server-enforced private/team/common visibility. Requires a running agent-memory-es instance.
Good to know before installing
Each user prompt (as a recall query), each user message and Hermes memory writes are sent to the agent-memory-es service at AMES_SERVICE_URL (default localhost:8123) with AMES_SERVICE_KEY; the model may store facts in shared team/common visibility tiers.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs AMES_SERVICE_URL, AMES_SERVICE_KEY), then asks whether to turn it on.
hermes plugins install agent_memory_es
Said no? Turn it on later:
hermes plugins enable agent_memory_es
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Kimchi Acp Provider
Community
The Kimchi coding harness (kimchi.dev) as a Hermes model provider over ACP stdio — spawns kimchi --mode acp keeping the harness's permission prompts on (YOLO opt-in via KIMCHI_ACP_ARGS="--mode acp --yolo"), the harness…
The Kimchi coding harness (kimchi.dev) as a Hermes model provider over ACP stdio — spawns kimchi --mode acp keeping the harness's permission prompts on (YOLO opt-in via KIMCHI_ACP_ARGS="--mode acp --yolo"), the harness owns its own auth, and completed tool calls surface as inline bullets in the reply.
Good to know before installing
Every turn spawns the local Kimchi CLI (`kimchi --mode acp`, overridable with KIMCHI_ACP_COMMAND / KIMCHI_ACP_ARGS) in Hermes's working directory and hands it the conversation. The harness calls kimchi.dev with its own stored login and runs its own shell, file and web tools. Its permission prompts stay on, and Hermes cancels them, so gated tools don't run, unless you opt in with KIMCHI_ACP_ARGS="--mode acp --yolo", which runs them with no approval prompt (not for shared gateway or cron use). To show setup status, it parses ~/.config/kimchi/config.json (or KIMCHI_CONFIG_PATH) only to check that a key is present; the key is never stored or sent by Hermes.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install kimchi-acp-provider
Said no? Turn it on later:
hermes plugins enable kimchi-acp-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Local fact-store memory provider: SQLite with FTS5 keyword search (word forms in English, Russian, Spanish, French, Portuguese, Indonesian, German, Hindi, Arabic, Bengali, Italian, Turkish, Polish and Dutch; Russian and English words of the same root at a lower weight, from bundled DerivBase.Ru (Apache-2.0) and Open English Wordnet (CC BY 4.0) pairs; German compounds and German and Spanish irregular verb forms from bundled Wikidata Lexemes tables (CC0); Chinese, Japanese, Korean and Thai by character n-grams, Chinese also split into words with jieba's dictionary (MIT), Traditional and Simplified Chinese compared as one through a bundled OpenCC table (Apache-2.0), Japanese spelling variants through SudachiDict's (Apache-2.0); synonyms through keywords the agent gives when it stores a fact), trust scores, entities with aliases and HRR (holographic reduced representation) recall, with no API key, network or model calls unless an optional online feature is switched on: search by meaning (embeddings: on) or rescoring by a cross-encoder (rerank: on), both off by default. The provider that shipped in-tree as plugins/memory/holographic, with the same provider name, config (plugins.hermes-memory-store) and memory_store.db, so `hermes update` moves existing users onto it; the schema change is additive and older versions keep working on the file. Taken over from Nous Research's handoff copy (NousResearch/hermes-plugin-holographic) and maintained by itpartypattaya; since 0.1.1 it fixes probe/reason/related/contradict, non-atomic writes and the never-written retrieval_count, gates recall by relevance without repeats, flags duplicates and conflicts when a fact is added, and honours the provider contract (read-only subagent contexts, hermes_home, mirror metadata, backup_paths); since 0.8.0 it keeps the history of every fact (change / end / correct / revert, search and list as_of a date) and `forget` erases a fact with all its versions from the file; since 0.16.0 a name guessed from a fact's text must be shaped like one and stays provisional until a second fact or a person confirms it, and entities can be renamed or deleted (the facts stay). Since 0.18.0 an optional, experimental second stage reorders the recall candidates with a cross-encoder at a /v1/rerank endpoint that answers in Cohere's shape without a key, such as llama-server (rerank: on, off by default). It does not raise hits: on the maintainer's held-out set (126 questions) it cut non-target facts in the top fi
Good to know before installing
Local only unless an optional online feature (search by meaning or rescoring) is switched on: reads and writes one SQLite file per profile ($HERMES_HOME/memory_store.db unless db_path says otherwise), or with scope: user / chat / thread one file per user or chat in holographic-users/ or holographic-chats/ next to it; each fact version records where it was written (platform, chat type, chat and thread id, session id, the turn author's id and display name, and whether the person said it or the agent inferred it; never chat titles), shown in full by list, history and export and not in recall unless recall_source: short (date, display name, this or another chat); each confirmation of a fact by a person keeps its time, session id and that person's id (no text); forget deletes that and the confirmations with the fact and keeps only an HMAC of the text, keyed by a random key stored in the same file, as a tombstone; delete_entity removes a name, its aliases and links from the entity index (never a fact) and keeps an HMAC of each name the same way (not of an alias only a fact's text gave, which goes without one), and both rewrite the file so freed pages keep no copy; the first writable open
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install holographic
Said no? Turn it on later:
hermes plugins enable holographic
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Kimchi Provider
Community
Kimchi (kimchi.dev) models via the OpenAI-compatible API — live catalog from the metadata endpoint, API-key auth, gateway unknown-model 400s classified as model_not_found, and an identifying User-Agent…
Kimchi (kimchi.dev) models via the OpenAI-compatible API — live catalog from the metadata endpoint, API-key auth, gateway unknown-model 400s classified as model_not_found, and an identifying User-Agent (hermes-agent/<version>) on every gateway request.
How to install
Install it. Hermes asks for any API keys it needs (this one needs KIMCHI_API_KEY), then asks whether to turn it on.
hermes plugins install kimchi-provider
Said no? Turn it on later:
hermes plugins enable kimchi-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Memtomem
Community
Markdown-first long-term memory shared across your coding agents: a local memtomem MCP server plus six memory skills (index, recall, remember, search, setup, status).
Markdown-first long-term memory shared across your coding agents: a local memtomem MCP server plus six memory skills (index, recall, remember, search, setup, status). Notes live in ~/.memtomem and are searchable from Hermes, Claude Code, Codex CLI and Kimi Code. Runs at user scope.
Good to know before installing
The server is launched with `uvx --python 3.12` and downloaded from PyPI (memtomem[onnx]==0.6.9) on first start, with Python 3.12 if uv cannot find one; it runs as a stdio MCP server for the session and reads the directories you ask it to index. With a fresh BM25-only configuration nothing else leaves the machine; an existing memtomem configuration may enable Hugging Face model downloads, a remote embedding or LLM provider, Cohere reranking or search webhooks. At startup the server also reads the nearest .env found upward from its installed location inside uv's cache (normally ~/.env), setting variables not already present, outside Hermes's MCP environment filtering.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install memtomem
Said no? Turn it on later:
hermes plugins enable memtomem
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
NetMind OpenAI-compatible model provider with a bundled model catalog.
Good to know before installing
Registers one model provider at import and nothing else (no tools, hooks, commands or threads); chat requests go to https://api.netmind.ai/inference-api/v1 through Hermes' own client with NETMIND_API_KEY, which Hermes reads; the plugin itself opens no connections, reads and writes no files and starts no processes.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs NETMIND_API_KEY), then asks whether to turn it on.
hermes plugins install netmind-provider
Said no? Turn it on later:
hermes plugins enable netmind-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Dense semantic memory for Hermes, stored in a Qdrant collection you host. Each turn becomes one point — user and assistant text verbatim, no LLM summarization step — embedded on-device at 384 dimensions with fastembed (ONNX; the default install is 206 MB of runtime dependencies with no torch, versus ~1.2 GB when sentence-transformers was a hard dependency) and recalled by cosine search with an optional session_id payload filter, so a result quotes what was actually said. Registers 7 tools: qdrant_search, qdrant_upsert, qdrant_recall, qdrant_collect, qdrant_prepare, qdrant_forget (point-targeted deletion, dry-run by default, no delete-all), and md_search (local markdown knowledge index). An optional recall-indicator display, display.level (off | summary | verbose, default off), adds latency and collection to the chat indicator at summary and one numeric line per operation to the log at verbose — numbers and identifiers only, never recalled message text or search-result snippets (the md_search line also echoes the query you searched — your own input, not recalled content); it is separate from the older top-level progress knob (off | minimal | verbose), which gates status events. sentence-transformers is selectable on GPU hosts; both backends produce identical vectors for all-MiniLM-L6-v2, so switching backends needs no re-embedding. A circuit breaker (5 consecutive I/O failures, 120 s cooldown) degrades to no new memories instead of crashing, and is_available()/check_backend()/unavailable_reason() distinguish a misconfiguration from a dead server. Optional hybrid dense+sparse RRF search (default off): one config key enables dense+BM25 RRF fusion. A fresh collection declares the sparse vector at creation and an existing dense-only store gets it added in place (Qdrant >= 1.18) with one bundled backfill command; measured +0.024 recall@10 on the LongMemEval 1,000-example benchmark. INT8 quantization is not implemented. Local embedded Qdrant mode is not supported: the provider is REST-only and needs a reachable server.
Good to know before installing
All memory content (verbatim user and assistant turns) is stored in and read back from the Qdrant server at the configured url (default http://localhost:6333); QDRANT_API_KEY is only needed for Qdrant Cloud, is kept in .env and read through the Hermes secret scope, and is never written to config.yaml or qdrant.json. Disclosure — the only other outbound traffic is a one-time download of the embedding model weights (plus the 10 MB Qdrant/bm25 sparse model when hybrid is on) from huggingface.co (fastembed falls back to storage.googleapis.com/qdrant-fastembed if that fails); embedding runs on-device and no memory content goes to any other party. Disclosure — settings are stored in <HERMES_HOME>/qdrant.json (created 0600, no credentials) and last-operation status in <HERMES_HOME>/qdrant-status.json. Disclosure — md_search returns snippets of local markdown to the model from ~/.hermes/skills, the Obsidian vault at ~/Documents/hermes_default, ~/.hermes/hermes-agent/website and any path in the md_docs_roots key of qdrant.json; its index is built only when you run scripts/md_ingest.py and lives in <base Hermes home>/state/md-search/ (shared by all profiles). Disclosure — the optional semant
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install qdrant
Said no? Turn it on later:
hermes plugins enable qdrant
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
ngrok AI Gateway
Community
The ngrok AI Gateway as a Hermes model provider (`--provider ngrok`, or ngrok AI Gateway in `hermes model`): one access key for all your model providers, with failover.
The ngrok AI Gateway as a Hermes model provider (`--provider ngrok`, or ngrok AI Gateway in `hermes model`): one access key for all your model providers, with failover. The model list comes live from your gateway, and gateway errors (unrouted model, access-policy and upstream-key failures, out of credit) map onto Hermes' fallback instead of a generic error.
Good to know before installing
Registers one model provider at import and nothing else (no tools, hooks, commands or threads); chat requests and the model list go to https://gateway.ngrok.ai/v1 (or NGROK_AI_GATEWAY_BASE_URL) through Hermes' own client with NGROK_AI_GATEWAY_KEY, which Hermes reads; the plugin itself opens no connections, reads and writes no files and starts no processes.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs NGROK_AI_GATEWAY_KEY), then asks whether to turn it on.
hermes plugins install ngrok-ai-gateway
Said no? Turn it on later:
hermes plugins enable ngrok-ai-gateway
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Git-backed Markdown memory with background curation and model-based recall.
Good to know before installing
Runs a background curation agent with terminal, file-read and patch tools on your configured auxiliary model, fed raw conversation turns; it commits and pushes the memory store to whatever git remotes you configure, and pulls shared submodules at session start; the engine runs via uv run, which fetches pyyaml==6.0.2 from PyPI; an optional dataset file (memory.stack.dataset_enabled, off by default) stores full transcripts locally with no rotation.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install stack
Said no? Turn it on later:
hermes plugins enable stack
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Runcrate
Community
Runcrate (runcrate.ai) open models via the OpenAI-compatible API at api.runcrate.ai/v1: live model list from /v1/models, DeepSeek V4.1 Flash as the default, and Hermes reasoning settings sent as reasoning_effort.
Runcrate (runcrate.ai) open models via the OpenAI-compatible API at api.runcrate.ai/v1: live model list from /v1/models, DeepSeek V4.1 Flash as the default, and Hermes reasoning settings sent as reasoning_effort.
Good to know before installing
Model provider; prompts, tool schemas and RUNCRATE_API_KEY go to api.runcrate.ai, and free requests can be served by third-party model providers under their own terms (runcrate.ai/privacy).
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs RUNCRATE_API_KEY), then asks whether to turn it on.
hermes plugins install runcrate-provider
Said no? Turn it on later:
hermes plugins enable runcrate-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Native XMemo memory provider for Hermes Agent — durable, identity-aware cross-session memory with semantic search.
Good to know before installing
Cloud memory provider: sends each user prompt (for recall and the search tools), model-written memories and mirrored built-in Hermes memory writes, plus a stable install ID, to xmemo.dev (Yonro Co., Ltd.); keeps a local SQLite cache of recalled results and a queue of unsent writes in $HERMES_HOME/xmemo_cache.db, partitioned by a hashed account identity; reads the key from XMEMO_KEY (also MEMORY_OS_API_KEY / MEMORY_OS_MCP_TOKEN) and, if none is set, falls back to the xmemo login token in ~/.config/xmemo/credentials.json (a catalog rule 11 deviation kept for existing users); workflow and destructive tools and timeline capture are off by default and opt-in.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs XMEMO_KEY), then asks whether to turn it on.
hermes plugins install xmemo
Said no? Turn it on later:
hermes plugins enable xmemo
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Safe AI Labs
Community
Safe AI Labs as a Hermes model provider (`--provider safeailabs`, or Safe AI Labs in `hermes model`): an OpenAI-compatible API with one endpoint per region, the model list from its /v1/models and the context window of…
Safe AI Labs as a Hermes model provider (`--provider safeailabs`, or Safe AI Labs in `hermes model`): an OpenAI-compatible API with one endpoint per region, the model list from its /v1/models and the context window of its fallback model declared.
Good to know before installing
Registers one model provider at import and nothing else (no tools, hooks, commands or threads); chat requests and the model list go to https://us.api.safeailabs.ai/v1 (or SAFE_AI_BASE_URL) through Hermes' own client with SAFE_AI_KEY, which Hermes reads; the plugin itself opens no connections, reads and writes no files and starts no processes.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs SAFE_AI_KEY), then asks whether to turn it on.
hermes plugins install safeailabs-provider
Said no? Turn it on later:
hermes plugins enable safeailabs-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Read-only Hermes Dashboard and Desktop plugin for inspecting built-in, holographic, Mem0, Honcho, Mnemosyne, Hindsight, and ByteRover memory.
Good to know before installing
Reads the active profile's memory files, provider configs and provider API keys (keys are never returned to the UI), and on user action sends search/recall/reflect queries with configured bank/peer/user IDs to the configured Mem0, Honcho and Hindsight endpoints (Hindsight Cloud default api.hindsight.vectorize.io) using those keys; ByteRover queries run the local `brv` CLI; for a local embedded Hindsight setup a contents read may start the `hindsight-embed` daemon; no memory is written.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install hermes-memory-ui
Said no? Turn it on later:
hermes plugins enable hermes-memory-ui
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Wallaby Token — curated open-weight inference (Kimi K3); adds Wallaby to the hermes model provider list.
How to install
Install it. Hermes asks for any API keys it needs (this one needs WALLABY_API_KEY), then asks whether to turn it on.
hermes plugins install wallaby-provider
Said no? Turn it on later:
hermes plugins enable wallaby-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
LoreConvo memory provider: cross-surface persistent session memory with search, recall and structured memory items.
Good to know before installing
Stores end-of-session transcripts (truncated, including tool output) and Hermes memory writes in ~/.loreconvo/sessions.db, a store shared across all Hermes profiles and other LoreConvo clients, and injects context from those sessions into each turn; a Pro license enables a one-time Hugging Face embedding-model download; installs the loreconvo package (BUSL-1.1) with lancedb, fastembed and onnxruntime into the Hermes environment.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs LORECONVO_HERMES_PROJECT), then asks whether to turn it on.
hermes plugins install loreconvo
Said no? Turn it on later:
hermes plugins enable loreconvo
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Agent Plugins v1 package: one skill plus one stdio MCP server (model_route) started with uv run from the checkout. Needs uv on PATH. Registers no Hermes tools or hooks; do not expect jev_* tools.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install jev-model-router
Said no? Turn it on later:
hermes plugins enable jev-model-router
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Vornik Companion
Community
Vornik as Hermes's hardened broker and long-term memory: Hermes holds no credentials; privileged work runs in operator-approved workflows on your own Vornik daemon.
Vornik as Hermes's hardened broker and long-term memory: Hermes holds no credentials; privileged work runs in operator-approved workflows on your own Vornik daemon.
Good to know before installing
Talks only to the self-hosted Vornik daemon at VORNIK_URL with VORNIK_BROKER_TOKEN (and VORNIK_MEMORY_TOKEN for memory); broker tools let the agent start operator-approved workflows on that daemon (writes are approved by a person in Vornik); when selected as memory provider it sends each turn's user query for recall, mirrors MEMORY.md and USER.md writes to the daemon, and asks the daemon to stop recalling (not delete) a mirrored entry that Hermes removes or replaces (/vornik-memory lists and /vornik-forget forgets what Vornik keeps); artifact tools expose only named outputs allowed by the approved workflow from completed runs and read bounded text or return short-lived binary download links from the daemon; task diagnosis tools (offered only when the daemon advertises companion-progress) return a broker task's execution events and closed error classes as the daemon serves them to a broker key; hermes vornik approvals on writes security.approval.transport (and vornik_companion.approval_namespace) to Hermes config, after which Hermes's own dangerous-command prompts (the redacted command text) are sent through the local vornikctl agent host-approval to the daemon and answered on the p
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs VORNIK_URL, VORNIK_BROKER_TOKEN, VORNIK_MEMORY_TOKEN), then asks whether to turn it on.
hermes plugins install vornik-companion
Said no? Turn it on later:
hermes plugins enable vornik-companion
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Account Switcher
Community
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…
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.
Good to know before installing
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.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install account-switcher
Said no? Turn it on later:
hermes plugins enable account-switcher
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Conxt
Community
Conxt governed memory provider — recall before each turn (personal, plus a team's approved memories), conversation capture in batches, and tools to search, remember, and propose team memories that a person approves in…
Conxt governed memory provider — recall before each turn (personal, plus a team's approved memories), conversation capture in batches, and tools to search, remember, and propose team memories that a person approves in Conxt before anyone recalls them.
Good to know before installing
Each turn's message (first 500 characters) and the conversation's user and assistant messages are sent to the Conxt API at mcp.conxt.dev (or an https CONXT_BASE_URL you set) with your CONXT_API_KEY; tool results are not sent, but text the assistant repeats is. Every entry Hermes's built-in memory tool adds is also copied to your Conxt memory, and conxt_remember / conxt_propose send the memory text they are given. No telemetry, no shell commands, and nothing is written automatically from cron or subagent runs. Requires a Conxt plan that includes API keys.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs CONXT_API_KEY), then asks whether to turn it on.
hermes plugins install conxt
Said no? Turn it on later:
hermes plugins enable conxt
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Adaptive Inference Routing
Community
Opt-in background scheduling for host-known delegated sessions, supported cron execution and explicit worker profiles using an existing Adaptive Inference gateway.
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.
Good to know before installing
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.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install adaptive-inference-routing
Said no? Turn it on later:
hermes plugins enable adaptive-inference-routing
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Identity
Community
View and edit the identity files (SOUL.md, MEMORY.md, USER.md, AGENT.md/HERMES.md) from chat, switch the active profile, and see per-line usage with colour-coded staleness timestamps.
View and edit the identity files (SOUL.md, MEMORY.md, USER.md, AGENT.md/HERMES.md) from chat, switch the active profile, and see per-line usage with colour-coded staleness timestamps. Adds six tools, /identity, /identity-switch and /identity-dashboard commands, and a utilization dashboard pane.
Good to know before installing
Reads and writes the identity files and a line-utilization record inside the selected profile directory under HERMES_HOME, and can change which profile is active; it performs no network I/O and holds no credentials. Disclosure — the model-callable identity_edit tool rewrites SOUL.md/MEMORY.md/USER.md/AGENT.md and identity_switch_profile writes HERMES_HOME/active_profile.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install identity
Said no? Turn it on later:
hermes plugins enable identity
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
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.
Good to know before installing
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.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install agent-hint
Said no? Turn it on later:
hermes plugins enable agent-hint
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Post Cortex
Community
Self-hosted long-term memory through a post-cortex daemon (https://github.com/julymetodiev/post-cortex): typed records (decisions, fixed problems, code changes) in an entity graph with local semantic search.
Self-hosted long-term memory through a post-cortex daemon (https://github.com/julymetodiev/post-cortex): typed records (decisions, fixed problems, code changes) in an entity graph with local semantic search. Recalls related records from the profile's workspace before each turn, puts the user card and the workspace's project card in the system prompt, mirrors built-in memory writes as records, and adds post_cortex_search, post_cortex_record and post_cortex_context (graph neighbours and reverse dependencies of named components) plus the post-cortex:memory skill. Records and captured turns carry the author hermes:<profile>. Profiles are mapped to shared workspaces by name glob, so a team of profiles recalls each other's records.
Good to know before installing
Talks only to the post-cortex daemon at the configured URL (default http://127.0.0.1:3737; a remote daemon over plain HTTP should sit on a private network or behind TLS): each non-trivial user prompt (up to 2,000 characters) is sent as a search query, and built-in memory writes and post_cortex_record calls are sent as records; with the opt-in capture setting (off by default) each completed turn of the primary agent is sent too: the user's message (with the sender's display name when Hermes knows it), the final reply, tool names and file paths, never tool output; nothing goes to third parties. Disclosure — the daemon stores records (and captured turns, deleted after its message_retention_days) in its own database and redacts credentials, e-mail addresses, phone numbers and public IPv4 addresses on write. Disclosure — PCX_TOKEN (and optionally PCX_URL) is read through the Hermes secret scope; the only local state is $HERMES_HOME/post-cortex.json (0600, never holds the token). No threads or processes of its own (capture runs on Hermes' sync worker), no shell commands, no deletes in post-cortex; cron and subagent runs never write automatically.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install post-cortex
Said no? Turn it on later:
hermes plugins enable post-cortex
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
AnyAPI.ai — unified OpenAI-compatible access to 400+ models; /model picker filtered to tool-capable routes.
Good to know before installing
Sends your prompts, tool results and ANYAPI_API_KEY to api.anyapi.ai, a third-party relay (AnyAPI Inc.) that forwards them to the upstream vendor serving the chosen model; billed to your AnyAPI account.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs ANYAPI_API_KEY), then asks whether to turn it on.
hermes plugins install anyapi
Said no? Turn it on later:
hermes plugins enable anyapi
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Local-first long-term memory backed by total-agent-memory (TAM) over MCP — saves each completed turn (with a noise filter; TAM's quality gate applies), recalls relevant memories before each turn within a token budget, and adds tam_recall / tam_save tools; optional TAM team server for shared company memory.
Good to know before installing
Local mode starts the separately installed tam binary as a child process per session with an allowlisted environment that carries no credentials: Hermes provider keys (including HF_TOKEN and OLLAMA_API_KEY), TAM_API_TOKEN and any *_TOKEN / *_API_KEY / *_SECRET / *_PASSWORD variable are withheld, and of HF_* / OLLAMA_* only non-secret settings pass (OLLAMA_URL, OLLAMA_EMBED_MODEL, HF_HOME, HF_HUB_OFFLINE); it stores user/assistant turn text plus built-in memory additions (not tool output) in TAM's store, by default outside HERMES_HOME and shared with other agents using TAM; remote mode sends the same text to the user-configured URL with an optional TAM_API_TOKEN bearer token; hermes memory setup tam writes tam.json, TAM_API_TOKEN to .env and memory.provider in config.yaml.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install tam
Said no? Turn it on later:
hermes plugins enable tam
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Read-only Home Assistant Recorder statistics tools (ha_list_statistics and ha_get_statistics).
Good to know before installing
Each call opens an authenticated WebSocket connection to the configured Home Assistant instance using the active profile HASS_TOKEN; returned statistic metadata and values become agent tool results and may reach the configured model provider. No telemetry, shell commands, background processes, credential storage or direct database access.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs HASS_TOKEN), then asks whether to turn it on.
hermes plugins install homeassistant-statistics
Said no? Turn it on later:
hermes plugins enable homeassistant-statistics
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Wicked Memory
Community
Persistent, wallet-scoped long-term memory for Hermes: remember, semantic recall, versioned update, history and approved hard delete, backed by the hosted Wicked Memory service, plus skills for using it and for…
Persistent, wallet-scoped long-term memory for Hermes: remember, semantic recall, versioned update, history and approved hard delete, backed by the hosted Wicked Memory service, plus skills for using it and for connecting the Wicked MCP server. Works with no setup (a wallet is generated on first use; search uses the service's shared free key).
Good to know before installing
Sends the memory text and search queries the agent chooses to memory.wickedapi.com (or your configured base_url), each request signed by the agent's wallet, plus one GET to /v1/status on first use to read the published community key; stores a generated private key unencrypted in the plugin data dir unless you supply WICKED_MEMORY_PRIVATE_KEY; no shell commands, no background processes, no telemetry; every wicked_memory_forget asks for human approval. The bundled wicked-suite-mcp skill walks you through adding the author's hosted Wicked MCP server (https://mcp.wickedapi.com/mcp, 75 tools, some x402-priced); that server is a separate manual step and is not covered by this pin.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install wicked-memory
Said no? Turn it on later:
hermes plugins enable wicked-memory
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Jev Effort Router
Community
Per-turn model AND reasoning-effort routing, decided by TypeSafe Jev, which rewrites the outgoing Ollama:Cloud provider request itself via the llm_request middleware.
Per-turn model AND reasoning-effort routing, decided by TypeSafe Jev, which rewrites the outgoing Ollama:Cloud provider request itself via the llm_request middleware. Every other catalog router picks a model and stops, and the two entries that touch reasoning effort move it by hand; this one decides both per turn. Ollama:Cloud only — no other provider is routed, and every failure path leaves the request byte-identical to a plugin-disabled run.
Good to know before installing
Sends the current user message and up to 4 previous turns to OpenRouter's Decisions API (typesafe/jev, billed to your OpenRouter key) on every routed turn.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs OPENROUTER_API_KEY), then asks whether to turn it on.
hermes plugins install jev-effort-router
Said no? Turn it on later:
hermes plugins enable jev-effort-router
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Gbrain Plugin
Community
Community (non-official) plugin wiring GBrain (garrytan/gbrain, MIT) into Hermes as a local memory layer over MCP: /gbrain status command, desktop starter, and per-profile Postgres setup/update/maintenance scripts.
Community (non-official) plugin wiring GBrain (garrytan/gbrain, MIT) into Hermes as a local memory layer over MCP: /gbrain status command, desktop starter, and per-profile Postgres setup/update/maintenance scripts. Not affiliated with upstream.
Good to know before installing
The plugin itself only adds a /gbrain status command and a Desktop pane; GBrain is installed by the user via scripts/setup.sh, which installs bun (curl|bash), an unpinned gbrain from GitHub HEAD, and provisions PostgreSQL with sudo.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install gbrain-plugin
Said no? Turn it on later:
hermes plugins enable gbrain-plugin
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Native Kiro model provider over HTTPS with AWS Builder ID / IAM Identity Center device login.
Good to know before installing
Dynamically registers an AWS OIDC public client named `hermes-kiro` during login and stores OAuth grants in Hermes' profile-scoped auth.json credential pool.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install kiro-provider
Said no? Turn it on later:
hermes plugins enable kiro-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Local Markdown memory provider with multilingual lexical recall and profile-local durable notes. Create-only writes are limited to primary agent contexts. Uses published birkin-mnemosyne 0.7.2 with compatible updates below 1.0; no API key, network service or telemetry. Session handoff capture is opt-in. Native Hermes memory files remain separate.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install birkin-mnemosyne
Said no? Turn it on later:
hermes plugins enable birkin-mnemosyne
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Liquid Inference Provider
Community
Liquid Inference, an auction exchange for LLM inference, as a Hermes model provider (provider id `liquid`): the exchange sends each request to the eligible offer with the lowest estimated charge, and each request…
Liquid Inference, an auction exchange for LLM inference, as a Hermes model provider (provider id `liquid`): the exchange sends each request to the eligible offer with the lowest estimated charge, and each request carries a cap on its charge (1 USD by default; set `model.default_headers.x-liquid-cap-usd` to change it). Submitted by Architect, the operator of Liquid Inference.
Good to know before installing
Registers one model provider at import and nothing else (no tools, hooks, commands or threads). Chat requests (full prompts and tool results), auxiliary tasks (model `liquid.auto`) and the model list go through Hermes's own client to https://router.inference.ai.exchange/v1 (or LIQUID_BASE_URL) with LIQUID_API_KEY. The exchange forwards each request to whichever third-party provider wins its auction. Each request is billed up to its `x-liquid-cap-usd` cap (1 USD by default, set with `model.default_headers.x-liquid-cap-usd`) and carries `X-Title: Hermes Agent`. A top-level `reasoning_effort` is sent only when you select a level.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs LIQUID_API_KEY), then asks whether to turn it on.
hermes plugins install liquid-inference-provider
Said no? Turn it on later:
hermes plugins enable liquid-inference-provider
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Local-only web dashboard for browsing and visualising Mnemosyne memories, triples, stats, and consolidation history.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install mnemosyne-dashboard
Said no? Turn it on later:
hermes plugins enable mnemosyne-dashboard
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Direct Mistral Vibe subscription access — browser PKCE login, dedicated credential pool, account-visible catalog discovery, relay-verified reasoning_effort for GLM models.
Good to know before installing
Unofficial, not affiliated with Mistral. "hermes auth add mistral-vibe" runs a browser PKCE sign-in against Mistral's Vibe console endpoints (console.mistral.ai/api/vibe/*) and stores the returned API key in the Hermes credential pool (auth.json, slice mistral-vibe), or uses MISTRAL_VIBE_API_KEY. Prompts and model listing go only to api.mistral.ai/v1. Vibe subscription coverage of API use is not guaranteed and may bill as pay-as-you-go. Logout clears the local pool but does not revoke the key remotely.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs (this one needs MISTRAL_VIBE_API_KEY), then asks whether to turn it on.
hermes plugins install mistral-vibe
Said no? Turn it on later:
hermes plugins enable mistral-vibe
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
Cognition
Community
Cognition session-end memory — facts regex-extracted from your messages at session end, stored in a local SQLite queue and JSON log under one repo label, with fetch-first merge, per-entry provenance, and budgeted…
Cognition session-end memory — facts regex-extracted from your messages at session end, stored in a local SQLite queue and JSON log under one repo label, with fetch-first merge, per-entry provenance, and budgeted prefetch (tools: cognition_search, cognition_status, read-only cognition_dream).
Good to know before installing
Saves verbatim excerpts (up to 400 chars) of your own messages that match preference/decision phrases in plain local files under HERMES_HOME/cognition/ (SQLite queue, JSON commit log) and re-injects them into later prompts sent to your model provider; the author name is read from ~/.gitconfig; the plugin makes no network calls of its own.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install cognition
Said no? Turn it on later:
hermes plugins enable cognition
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Per-call delegation across your own subscriptions. The model sends each subagent to a named route with delegate_to(route, goal, context); each route pins a provider Hermes already knows (Claude, Codex or Antigravity subscription plugins, ACP provider plugins) and a model. delegate_task, by contrast, uses the one target in config.yaml. Subagents launch through ctx.subagent_lifecycle. A child counts as out of quota only on an unambiguous provider error or a measured window at 100 %, and the route never switches on its own: a policy decides whether the parent asks you, picks another route, or waits.
Good to know before installing
A pre_tool_call hook blocks delegate_task spawns while routes are configured (list/steer/stop pass); an api_request_error hook records quota errors in memory. Quota is read with Hermes' fetch_account_usage (Anthropic and OpenAI Codex) on /ohx quota and after a provider error. No other network egress, no files written.
From the plugin's catalog listing.
How to install
Install it. Hermes asks for any API keys it needs, then asks whether to turn it on.
hermes plugins install orquehelx
Said no? Turn it on later:
hermes plugins enable orquehelx
Restart Hermes so it loads the plugin.
Some plugins also need a sign-in or settings: see Full setup & settings. Memory plugins ask whether to become your memory provider instead. Steps follow the official plugins guide↗ (opens in a new tab).
Full setup steps, settings and screenshots are on the plugin's official catalog page from Nous Research.
No plugins match that search.
Read before you install
Plugins run code inside your agent. Official plugins come from Nous Research's own repo. Community plugins are independent projects, so check the source, the permissions they ask for and any API keys they need. New to plugins? Start with skills vs plugins vs MCP.
