Skip to content

Commit d66cecf

Browse files
Merge pull request #5 from OnlyTerp/devin/1776393615-v0-10-refresh
- Add Parts 12-16: Web Dashboard, Nous Tool Gateway, Fast Mode + background watchers + pluggable context engine, new messaging platforms (iMessage, WeChat, Android/Termux), backup/import + /debug + security hardening - README: What's New section covering both v0.9.0 'everywhere' and v0.10.0 'tool gateway' releases, updated TOC to 16 parts, expanded provider table (Nous Portal, native xAI/Xiaomi, Kimi, z.ai, MiniMax, Arcee, HF), Android/Termux install, dashboard quick start - Part 4: reframe Telegram as one of 16 gateway platforms with cross-references to the new adapters - Part 9: native adapter matrix, Nous Portal OAuth flow, xAI live search + MiMo reasoning modes, new provider config examples Co-authored-by: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
2 parents 10abd24 + cd58f31 commit d66cecf

8 files changed

Lines changed: 1609 additions & 23 deletions

README.md

Lines changed: 61 additions & 16 deletions
Original file line numberDiff line numberDiff line change
@@ -1,17 +1,44 @@
11
# Hermes Optimization Guide
22

3-
> **Tested on Hermes Agent (latest) — April 2026** · 11 parts · Battle-tested on a live production deployment
3+
> **Tested on Hermes Agent v0.10.0 (v2026.4.16)** · 16 parts · Battle-tested on a live production deployment
44
5-
### Make Your Hermes Agent Actually Work — Setup, Migration, Knowledge Graphs, Telegram, Skills, Memory, Models, and Recovery
6-
#### Full setup walkthrough, OpenClaw migration, LightRAG graph RAG, Telegram bot integration, on-the-fly skill creation, memory architecture, custom models, and crash recovery
5+
### Make Your Hermes Agent Actually Work — Setup, Migration, Knowledge Graphs, Messaging, Skills, Memory, Models, Dashboard, Tool Gateway, and Recovery
6+
#### Full setup walkthrough, OpenClaw migration, LightRAG graph RAG, 16 messaging platforms, on-the-fly skill creation, memory architecture, custom models, the new local web dashboard, Nous Tool Gateway, Fast Mode, backup/import, and crash recovery
77

88
*By Terp - [Terp AI Labs](https://x.com/OnlyTerp)*
99

1010
---
1111

12+
## What's New (April 10–17, 2026)
13+
14+
Two major Hermes releases dropped this week. This guide is current as of both.
15+
16+
### [v0.9.0 — 2026.4.13 — "Everywhere"](https://github.com/NousResearch/hermes-agent/releases/tag/v2026.4.13)
17+
18+
- **Local web dashboard** (`hermes dashboard`) — full browser UI for config, API keys, sessions, logs, analytics, cron, and skills. See [Part 12](./part12-web-dashboard.md).
19+
- **Fast Mode** (`/fast`) — priority-tier inference on OpenAI & Anthropic, available on every gateway platform, not just the CLI. See [Part 14](./part14-fast-mode-watchers.md).
20+
- **Three new messaging adapters** — iMessage (via BlueBubbles), WeChat/Weixin, and WeCom (Enterprise WeChat). Brings the total to **16 platforms**. See [Part 15](./part15-new-platforms.md).
21+
- **Android / Termux** tested install path — run the full Hermes CLI on your phone. See [Part 15](./part15-new-platforms.md#android--termux-running-hermes-on-your-phone).
22+
- **Background process monitoring** (`watch_patterns`) — real-time regex event hooks on long-running processes, no more polling. See [Part 14](./part14-fast-mode-watchers.md#background-process-monitoring-watch_patterns).
23+
- **Pluggable context engine** — swap in a custom engine to filter memory, inject domain context, or pre-summarize tool output. See [Part 14](./part14-fast-mode-watchers.md#pluggable-context-engine).
24+
- **`hermes backup` / `hermes import`** — first-class portable backups with interactive conflict resolution. See [Part 16](./part16-backup-debug.md).
25+
- **Dashboard plugins** — third-party tabs that extend the web UI. See [Part 12](./part12-web-dashboard.md#dashboard-plugins-extend-the-ui).
26+
- **Proxy support, webhook secret validation, SSRF protection, log redaction** — hardening pass across every adapter. See [Part 16](./part16-backup-debug.md#security-hardening-v09--v010-notes).
27+
28+
### [v0.10.0 — 2026.4.16 — "Tool Gateway"](https://github.com/NousResearch/hermes-agent/releases/tag/v2026.4.16)
29+
30+
- **Nous Tool Gateway** — paid [Nous Portal](https://portal.nousresearch.com) subscribers get web search (Firecrawl), image generation (FAL FLUX 2 Pro), TTS (OpenAI), and browser automation (Browser Use) through their subscription — **no extra API keys**. Per-tool `use_gateway: true` with clean precedence over direct keys. See [Part 13](./part13-tool-gateway.md).
31+
- **Native xAI and Xiaomi MiMo providers** — both now have first-class adapters with provider-specific features (Grok's live-X search, MiMo's reasoning modes), not just OpenRouter pass-through. See [Part 9](./part9-custom-models.md).
32+
- **`/compress <topic>`** — guided compression that preserves detail relevant to a topic. See [Part 14](./part14-fast-mode-watchers.md#compress-topic--guided-compression).
33+
- **`/debug` + `hermes debug share`** — single-command diagnostic bundler with upload endpoint for bug reports. See [Part 16](./part16-backup-debug.md#debug-and-hermes-debug-share).
34+
- **Approval bypass inheritance for subagents** — subagents inherit the parent session's approval posture; override per delegation. See [Part 16](./part16-backup-debug.md#approval-bypass-for-trusted-subagents).
35+
- `HERMES_ENABLE_NOUS_MANAGED_TOOLS` env flag **removed** — replaced by clean subscription detection + per-tool `use_gateway`. `hermes upgrade` migrates automatically.
36+
37+
---
38+
1239
## Table of Contents
1340

14-
1. [Setup](#part-1-setup-stop-fumbling-with-installation) — Install Hermes, configure your provider, first-run walkthrough
41+
1. [Setup](#part-1-setup-stop-fumbling-with-installation) — Install Hermes, configure your provider, first-run walkthrough (with Android/Termux)
1542
2. [SOUL.md Personality](#soulmd--give-your-agent-a-personality) — The Molty prompt, what good personality rules look like, how to fix a bland agent
1643
3. [OpenClaw Migration](#part-2-openclaw-migration-dont-leave-your-knowledge-behind) — Move your OpenClaw data, config, skills, and memory into Hermes
1744
4. [LightRAG — Graph RAG](#part-3-lightrag--graph-rag-that-actually-works) — Set up a knowledge graph that actually understands relationships, not just text similarity
@@ -20,9 +47,14 @@
2047
7. [Context Compression](./part6-context-compression.md) — Fix the silent context loss bug, configure compression thresholds, survive long sessions
2148
8. [Memory System](./part7-memory-system.md) — The three-tier memory architecture: persistent facts, conversation recall, procedural memory
2249
9. [Subagent Patterns](./part8-subagent-patterns.md) — Orchestrator/worker delegation, ACP subagents, parallel task execution
23-
10. [Custom Model Providers](./part9-custom-models.md) — Cerebras, Fireworks, Ollama, model aliases, fallback chains
50+
10. [Custom Model Providers](./part9-custom-models.md) — Cerebras, Fireworks, Ollama, native xAI/MiMo/z.ai/Kimi/MiniMax/Arcee, Nous Portal, model aliases, fallback chains
2451
11. [SOUL.md Anti-Patterns](./part10-soul-antipatterns.md) — What makes an agent annoying vs useful, the formula that works
2552
12. [Gateway Recovery](./part11-gateway-recovery.md) — Crash detection, auto-recovery, common failure modes, health checks
53+
13. [Web Dashboard](./part12-web-dashboard.md)**New.** `hermes dashboard`, the full browser UI — config, keys, sessions, logs, analytics, cron, skills, REST API, plugins
54+
14. [Nous Tool Gateway](./part13-tool-gateway.md)**New.** Web search, image gen, TTS, and browser automation through a single Nous Portal subscription
55+
15. [Fast Mode & Background Watchers](./part14-fast-mode-watchers.md)**New.** `/fast` priority tier, `watch_patterns` real-time process monitoring, pluggable context engine, `/compress <topic>`
56+
16. [New Platforms (iMessage, WeChat, Android)](./part15-new-platforms.md)**New.** BlueBubbles/iMessage, Weixin/WeCom, Android via Termux — the full 16-platform lineup
57+
17. [Backup, Import & `/debug`](./part16-backup-debug.md)**New.** Portable `hermes backup`/`import`, `/debug` bundler, `hermes debug share`, security hardening
2658

2759
---
2860

@@ -52,10 +84,11 @@ After this guide:
5284

5385
## Prerequisites
5486

55-
- A Linux/macOS machine (or WSL2 on Windows)
87+
- A Linux/macOS machine (or WSL2 on Windows, or **Android via Termux** — see [Part 15](./part15-new-platforms.md#android--termux-running-hermes-on-your-phone))
5688
- Python 3.11+ and Git
57-
- An API key for at least one LLM provider (Anthropic, OpenAI, OpenRouter, etc.)
89+
- An API key for at least one LLM provider (Anthropic, OpenAI, OpenRouter, Nous Portal, etc.)
5890
- Optional: Ollama for local embeddings (free vector search)
91+
- Optional: A paid [Nous Portal](https://portal.nousresearch.com) subscription to use the [Tool Gateway](./part13-tool-gateway.md) — web search, image gen, TTS, and browser automation with no extra keys
5992

6093
---
6194

@@ -83,16 +116,21 @@ LLM Provider (Claude, GPT, local models)
83116
## Quick Start
84117

85118
```bash
86-
# 1. Install Hermes
119+
# 1. Install Hermes (Linux/macOS/WSL2/Android)
87120
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
88121

89-
# 2. Configure
122+
# 2. Configure providers and tools
90123
hermes setup
91124

92-
# 3. Start chatting
125+
# 3a. Start chatting in the terminal
93126
hermes
127+
128+
# 3b. Or launch the new browser dashboard (v0.9+)
129+
hermes dashboard
94130
```
95131

132+
The dashboard is the fastest way to configure everything without touching YAML. See [Part 12](./part12-web-dashboard.md) for the full tour.
133+
96134
For the full walkthrough including optimization, read each part in order.
97135

98136
---
@@ -118,6 +156,8 @@ curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scri
118156
119157
> **Windows users:** Native Windows is not supported. Install [WSL2](https://learn.microsoft.com/en-us/windows/wsl/install) and run the command from inside WSL. It works perfectly.
120158
159+
> **Android users (new in v0.9):** the same installer detects Termux and installs the tested `[termux]` extra bundle automatically — CLI, cron, PTY/background terminal, Telegram gateway, MCP, Honcho, ACP. See [Part 15 — Android / Termux](./part15-new-platforms.md#android--termux-running-hermes-on-your-phone).
160+
121161
### What the Installer Does
122162
123163
The installer handles everything automatically:
@@ -156,15 +196,20 @@ Supported providers and recommended models:
156196

157197
| Provider | Top Models | Best For | Env Variable |
158198
|----------|-----------|----------|-------------|
159-
| **Anthropic** | Opus 4.6, Sonnet 4 | Best reasoning, complex tasks, coding | `ANTHROPIC_API_KEY` |
160-
| **OpenAI** | GPT-5.4 Pro, o3, GPT-4.1 | Strong tool use, fast inference, huge context | `OPENAI_API_KEY` |
161-
| **Xiaomi** | MiMo V2 Pro | Fast, cheap, great for orchestration and routing | `OPENROUTER_API_KEY` |
199+
| **Nous Portal** | Hermes 5, Hermes 4 405B | Built-in [Tool Gateway](./part13-tool-gateway.md) — web search/image/TTS/browser with no extra keys | Auth via `hermes model` |
200+
| **Anthropic** | Opus 4.6, Sonnet 4 | Best reasoning, complex tasks, coding, `/fast` priority tier | `ANTHROPIC_API_KEY` |
201+
| **OpenAI** | GPT-5.4 Pro, o3, GPT-4.1 | Strong tool use, fast inference, huge context, `/fast` priority tier | `OPENAI_API_KEY` |
202+
| **Xiaomi MiMo** | MiMo V2 Pro *(native adapter)* | Fast, cheap, native reasoning modes, great for orchestration | `XIAOMI_API_KEY` |
203+
| **xAI** | Grok 3, Grok 3 Mini *(native adapter)* | Fast, good reasoning, native live-X search | `XAI_API_KEY` |
204+
| **Kimi / Moonshot** | Kimi 2.5 | Big context, excellent for entity extraction / LightRAG ingestion | `MOONSHOT_API_KEY` |
205+
| **z.ai / GLM** | GLM-5, GLM-5 Air | Strongest open-weights model, great for translation + tools | `ZAI_API_KEY` |
162206
| **Google** | Gemini 3.1 Pro | Massive context (2M tokens), multimodal, cheap | `GEMINI_API_KEY` |
163-
| **MiniMax** | M2.7 | Good balance of speed and quality | `OPENROUTER_API_KEY` |
164-
| **xAI** | Grok 3, Grok 3 Mini | Fast, good reasoning, real-time X/Twitter access | `XAI_API_KEY` |
207+
| **MiniMax** | M2.7 | Good balance of speed and quality | `MINIMAX_API_KEY` |
165208
| **Cerebras** | Llama 4 Scout, Qwen 3 32B | Blazing fast inference (2000+ tok/s), cheap | `CEREBRAS_API_KEY` |
166209
| **Groq** | Llama 4, Qwen 3 | Very fast inference, limited context | `GROQ_API_KEY` |
167-
| **OpenRouter** | All of the above + 100 more | Access every model from one key, auto-fallback | `OPENROUTER_API_KEY` |
210+
| **Arcee** | AFM-4.5, Caller | Function-calling specialists, cheap | `ARCEE_API_KEY` |
211+
| **Hugging Face** | Any TGI/TEI endpoint | Self-hosted and Inference Endpoints | `HF_TOKEN` |
212+
| **OpenRouter** | All of the above + 200 more | Access every model from one key, auto-fallback | `OPENROUTER_API_KEY` |
168213
| **Ollama** (local) | Qwen 3.5 Opus Distilled V3, Gemma 4, Nemotron | Free, private, runs on your GPU — great for embeddings and simple tasks | None needed |
169214

170215
### Local Models (Ollama)

0 commit comments

Comments
 (0)