MemFlow Relay
Research Preview — MemFlow Relay is currently in research preview. APIs and behaviours may change without prior notice.
MemFlow Relay is a local HTTP proxy for Gemini API requests, developed by LamoonAGI. It transparently intercepts your Gemini calls and — depending on the configured operating mode — can analyse, optimise, or shadow-test context window usage before forwarding the request to Google Gemini.
Features
- Shadow mode — observe without modification; original request is always forwarded unchanged.
- Safe mode — apply optimizations only when confidence is high; fall back to original on any doubt.
- Optimize mode — apply MemFlow context optimizations and track token savings.
- Fail-open default — if the MemFlow optimizer is unreachable or times out, the relay always forwards the original request to Gemini.
- Credential isolation —
GEMINI_API_KEYnever leaves your machine; it is never sent to LamoonAGI or MemFlow cloud services. - Secret redaction — all logs, exceptions, and telemetry automatically strip API keys, Bearer tokens, and cookies.
- Telemetry modes —
off,metrics_only(default, no prompt content), ordiagnostic.
Installation
pip install memflow-relay
Requires Python ≥ 3.11.
Quick Start
# Set credentials
export MEMFLOW_API_KEY="lago_your_key_here"
export GEMINI_API_KEY="your_gemini_key_here"
# Start relay in shadow mode (safe default)
memflow-relay serve
# Verify setup
memflow-relay doctor
# Show live stats
memflow-relay stats
Your Gemini SDK base URL should be set to http://127.0.0.1:8765.
CLI Commands
| Command | Description |
|---|---|
memflow-relay serve |
Start the local proxy (default port 8765) |
memflow-relay doctor |
Verify credentials, connectivity, and port availability |
memflow-relay test |
Run end-to-end shadow mode verification |
memflow-relay stats |
Display analyzed / optimized / bypassed counts and token savings |
memflow-relay benchmark |
Run baseline vs optimized Gemini workload comparison |
memflow-relay version |
Report package version and research preview status |
Configuration
All configuration is via environment variables. See .env.example for a full reference.
| Variable | Default | Description |
|---|---|---|
MEMFLOW_API_KEY |
— | LamoonAGI API key (required for optimize/safe modes) |
GEMINI_API_KEY |
— | Google Gemini API key (never sent to LamoonAGI) |
MEMFLOW_API_URL |
https://api.lamoonagi.com |
MemFlow optimizer endpoint |
MEMFLOW_MODE |
shadow |
Operating mode: shadow, safe, optimize |
MEMFLOW_TELEMETRY |
metrics_only |
Telemetry level: off, metrics_only, diagnostic |
MEMFLOW_PORT |
8765 |
Local proxy port |
MEMFLOW_FAIL_OPEN |
true |
Forward original request if optimizer fails |
Security
See SECURITY.md for vulnerability disclosure policy.
Key guarantee: GEMINI_API_KEY is used exclusively to authenticate requests to generativelanguage.googleapis.com. It is never included in requests to api.lamoonagi.com or any other endpoint.
License
MIT — see LICENSE.
© 2024 LamoonAGI. MemFlow is a product of LamoonAGI. Powered by LamoonAGI Context VMM technology.
Metadata
Release files for memflow-relay 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| memflow_relay-0.1.0.tar.gz | 25.8 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| memflow_relay-0.1.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 49.3 kB
Release files / memflow_relay-0.1.0.tar.gz
| Download URL | memflow_relay-0.1.0.tar.gz |
|---|---|
| Size | 25.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
8658553e2a41e130245bd9f0a1e16a9af5eb1def45747cd6896ce9d3a068d3e6
|
|
BLAKE2b-256 checksum How to use checksums |
5ba7d5e584578870e39ac6c93ae6e2d3b9e315ee099d2feee0ac71d971191a52
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency logRelease files / memflow_relay-0.1.0-py3-none-any.whl
| Download URL | memflow_relay-0.1.0-py3-none-any.whl |
|---|---|
| Size | 23.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
40d0959746bba824439024aedafb54116676e394abaeab2253f2e49752a82826
|
|
BLAKE2b-256 checksum How to use checksums |
87154310b2d62e68e39d394ffb63ef7f38fd556466f61f2a38734ac40a01a1ab
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency log