hermes-gpt
Enables MCP-based interaction with a local Hermes Agent installation, providing file read/search, skill management, and optional sandboxed write, terminal, and memory tools.
README
hermes-gpt

hermes-gpt is a standalone MCP sidecar for Hermes Agent. It imports selected local Hermes Agent internals at runtime and exposes them to MCP clients without modifying Hermes Agent source files.
This is a local-dev release. It is not a hosted service, not a fork of Hermes Agent, not a generic remote dev container, and not a replacement for DevSpace.
Security posture
By default, hermes-gpt is designed for a trusted local machine:
- HTTP binds to
127.0.0.1by default. - Tools advertise
noauthso ChatGPT can use Developer Mode withNo Authentication; optional bridge auth is enforced by the HTTP endpoint itself. - Public HTTP can be protected with
HERMES_GPT_HTTP_TOKENusing query parameters or an HTTP Bearer token header. - Reads are limited by
HERMES_GPT_READ_ROOTSand writes are confined toHERMES_GPT_SAFE_ROOT. - Write is create-only; patch, terminal execution, memory writes, and session search are disabled or hidden by default.
Do not expose this server publicly without HERMES_GPT_HTTP_TOKEN or an equivalent outer authentication layer. Prefer a stable Cloudflare named tunnel or reserved ngrok domain over quick trycloudflare.com URLs when ChatGPT connector URLs must survive restarts.
Prerequisites
- Python 3.10+
- A local Hermes Agent install
- MCP Python SDK and Uvicorn
Install dependencies:
cd ~/hermes-gpt
python -m pip install -r requirements.txt
Local MCP clients
Stdio mode is for local MCP clients that support subprocess MCP servers:
cd ~/hermes-gpt
python server.py
Example client command:
{
"command": "python",
"args": ["C:\\Users\\asimo\\hermes-gpt\\server.py"]
}
Local HTTP
HTTP mode uses FastMCP streamable HTTP:
cd ~/hermes-gpt
python server.py --http --host 127.0.0.1 --port 7677
Local endpoint:
http://127.0.0.1:7677/mcp
If you bind to anything other than loopback in the default local-dev profile, the server prints a warning. This warning means the configuration is not release-safe.
ChatGPT local testing
ChatGPT developer mode expects a remote MCP endpoint. Do not enter a localhost URL such as http://127.0.0.1:4750; ChatGPT fetches the MCP configuration through its connector path, where 127.0.0.1 is not your machine.
For short local testing only:
cd C:\Users\asimo\hermes-gpt
python server.py --http --host 127.0.0.1 --port 4750
In another terminal:
& "C:\Program Files (x86)\cloudflared\cloudflared.exe" tunnel --url http://127.0.0.1:4750 --http-host-header 127.0.0.1:4750
For token-protected local testing, start the server with a stable bridge token:
HERMES_GPT_HOST_IDENTITY=Dell \
HERMES_GPT_HTTP_TOKEN=placeholder-token \
HERMES_GPT_SAFE_ROOT=/path/to/sandbox \
HERMES_GPT_READ_ROOTS=/path/to/read/root \
HERMES_GPT_ENABLE_WRITE=1 \
python server.py --http --host 127.0.0.1 --port 4750
In ChatGPT, configure:
- Protocol: Streaming HTTP
- MCP server URL:
https://<your-host>/mcp?hermes_token=placeholder-token - Authentication: No Authentication
The token is intentionally part of the server URL because many ChatGPT connector flows cannot set custom headers. Do not publish real connector URLs. Remove and recreate the connector if ChatGPT cached older tool metadata.
Tool gates
Default visible tools:
hermes_server_config()— host identity, read roots, write root, auth status, and safety modes without revealing token values.hermes_self_test()— local bridge diagnostics for roots, token, write sandbox, and sensitive-path blocking.hermes_tree(path=".", depth=2, limit=100)— shallow directory tree.hermes_read_file(path, offset=1, limit=500)hermes_search_files(pattern, target="content", path=".", file_glob=None, limit=50);target="path"is accepted as a filename-search alias.hermes_list_files(path=".", limit=50)hermes_view_image(path)— returns PNG/JPEG/WebP/GIF as MCP image content.hermes_skill_list()hermes_skill_view(name)
Opt-in tools and actions:
| Capability | Env var | Default |
|---|---|---|
| Create-only sandbox write | HERMES_GPT_ENABLE_WRITE=1 + HERMES_GPT_SAFE_ROOT=/path/to/sandbox |
Hidden |
Memory add, replace, remove |
HERMES_GPT_ENABLE_MEMORY_WRITE=1 |
Disabled |
| Session search | HERMES_GPT_ENABLE_SESSION_SEARCH=1 |
Hidden |
| Terminal command execution | HERMES_GPT_ENABLE_TERMINAL=1 |
Hidden |
Terminal timeout is capped at 120 seconds even when enabled.
CodexPro-inspired bridge knobs
| Setting | Purpose |
|---|---|
HERMES_GPT_HOST_IDENTITY |
Human label such as Dell or Pedro returned by hermes_server_config. |
HERMES_GPT_HTTP_TOKEN |
Enables query/Bearer token enforcement on HTTP/SSE transports. |
HERMES_GPT_REQUIRE_HTTP_TOKEN=1 |
Reject HTTP traffic if no token is configured. |
HERMES_GPT_READ_ROOTS |
os.pathsep-separated read-only roots visible to read/search/tree/image tools. |
HERMES_GPT_PATH_ALIASES |
Public-to-local path aliases, e.g. /linus1=/home/imac-hermes/linus1/shared. |
HERMES_GPT_SAFE_ROOT |
The only directory where write tools may create files. |
See HERMESPRO_PLAN.md for Dell/Pedro deployment examples and the hybrid design rationale.
Remote profile
--profile remote is intended for public/tunneled testing. Configure HERMES_GPT_HTTP_TOKEN before exposing it through a public URL:
HERMES_GPT_HTTP_TOKEN=placeholder-token python server.py --http --profile remote --host 127.0.0.1 --port 7677
For temporary no-auth experiments only, you can bypass the remote-profile guard with both:
HERMES_GPT_UNSAFE_REMOTE_NOAUTH=1
python server.py --http --profile remote --i-understand-this-is-unsafe
Do not use the no-auth bypass for release.
Release checklist
Before publishing:
- No
*.pemfiles. - No
*.logor*.err.logfiles. - No
__pycache__/or*.pyc. python -m py_compile server.pypasses.pytestpasses.- Server binds to loopback by default.
- Terminal, write tools, memory writes, and session search are disabled by default.
Current capability notes
The feasibility probe passed in this environment:
- Hermes source root:
C:\Users\asimo\AppData\Local\hermes\hermes-agent - File tools: available
- Terminal tool: available, gated by
HERMES_GPT_ENABLE_TERMINAL=1 - Memory tool: available
- Skill discovery: available through local and bundled skill directories
- Session search: available through
SessionDB.search_messages - FastMCP stdio: available
- FastMCP streamable HTTP: available
See FEASIBILITY.md for probe details and exact signatures.
License
MIT. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.