heropen
Give your AI agent long-term memory. Data stays on your machine; search costs zero tokens.
Why "heropen"
The name comes from two places: her from Hermes (the agent you are reading this with), open from OpenClaw (openness). her + open put together is heropen.
Written out, heropen starts with hero — evoking the Marvel superhero trope. It is also a companion that remembers you and writes things down for you.
Install
pip install heropen
Restart your agent. That's it.
On first launch it auto-detects your agent (Claude Code, Cursor, Windsurf, or any MCP client), sets up the database, and registers the memory tools. Your agent will notice the new install and walk you through setup.
30-second quickstart
# Save a memory
heropen add "Project uses FastAPI + SQLAlchemy, tests with pytest"
# Search memories
heropen search "project tech stack"
# Check status
heropen status
# Diagnose issues
heropen diagnose
Connect your agent (MCP)
Works with any MCP-compatible agent. v1.8+ auto-detects and configures — no manual steps.
Or add it manually to your agent config:
{
"mcpServers": {
"heropen": {
"command": "heropen",
"args": ["mcp"]
}
}
}
Restart your agent and it has memory. Store a bug fix once, remember it permanently across sessions.
Privacy promise
Data stays on your machine. No telemetry. No heartbeat pings. All memory is stored in a local SQLite database. Vector search uses a local embedding model by default (fastembed, pip install heropen[embedding]) — fully offline, zero cost. Optionally, you can point it at your own self-hosted embedding endpoint by setting the EMBEDDING_ENDPOINT and EMBEDDING_API_KEY environment variables (OpenAI-compatible /v1/embeddings), so no third-party cloud is ever billed. Memory text is only used to generate vectors and is never reported.
If neither a local embedding model nor a self-hosted endpoint is configured, search automatically degrades to fast full-text (FTS) matching — still fully offline and zero cost. So pip install heropen works with zero setup; embeddings only upgrade search quality, they never gate basic use.
Open-source scope
The free edition is fully open source (Apache-2.0). The commercial layer (Plus / Enterprise) is closed source.
Why heropen
| heropen (free) | other solutions | |
|---|---|---|
| Storage | unlimited | usually capped |
| Searches | unlimited | pay per query |
| Needs network | no | yes |
| Data ownership | your machine | their servers |
| Install | one pip install |
server + config |
Free = full core features. No crippled functionality.
Links
- Home: heropen.net
- Docs: heropen.net/heropen/docs
- GitHub: github.com/Koradji77/heropen
License
Apache-2.0
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file heropen-1.8.7.tar.gz.
File metadata
- Download URL: heropen-1.8.7.tar.gz
- Upload date:
- Size: 206.5 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
88f36ccb34f4ff27d8bc697f18f4d1240349d18b343d91165ce1f6a20459b2f3
|
|
| MD5 |
8dc46295c3a0d75a92103814c525ea1a
|
|
| BLAKE2b-256 |
e33efe71d51f07b97f1a4947ba56f592cde425bad5e865c6cf2d17257c1c103b
|
File details
Details for the file heropen-1.8.7-py3-none-any.whl.
File metadata
- Download URL: heropen-1.8.7-py3-none-any.whl
- Upload date:
- Size: 77.8 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.2.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
14ef8ca27d62a32878e388ea6829045d54f79e2a1e6efa78f482d078b796302c
|
|
| MD5 |
8ded15f1387caab6107d9f15fae4a72c
|
|
| BLAKE2b-256 |
6d145a375489a124375f1600766356f27072b5c6fa3e0ec99587a4a46c743b98
|
