Analysis updated 2026-05-18
Give a coding agent lasting memory of project conventions and past incidents.
Build a personal assistant that remembers preferences and appointments privately.
Run an ops agent that recalls which hosts are flaky and what past incidents looked like.
Keep an agent working offline even when cloud summarization services are down.
| augmem/cortext-hermes-plugin | 0xallam/my-recipe | 0xhassaan/nn-from-scratch | |
|---|---|---|---|
| Stars | 0 | — | 0 |
| Language | Python | Python | Python |
| Last pushed | — | 2022-11-22 | — |
| Maintenance | — | Dormant | — |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 2/5 | 2/5 | 4/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires two Hermes CLI commands and a restart, the clone is large (~135MB) since it bundles the local model.
This project gives the Hermes Agent a memory that survives after a chat session ends. Normally, once you close a Hermes session, the agent forgets everything you told it. This plugin fixes that by connecting Hermes to Cortext, a memory engine that runs entirely on your own computer and stores what it learns in a single SQLite database file. Once installed, you tell Hermes something once, like which branch you deploy from or an allergy you have, and it remembers that fact in future sessions, even brand new ones, without you repeating yourself or keeping a separate notes file. When you later correct a fact, the old version is replaced instead of being kept alongside the new one, so the agent does not get confused by outdated information. The plugin also replaces Hermes's built in way of handling long conversations. Normally, when a chat gets too long, Hermes calls another AI model to summarize and compress the history, which can fail if that outside service is down. This plugin does that compression locally and near instantly instead, so it keeps working even without an internet connection. Privacy is a core feature here. Everything is stored in one file on your machine, there are no calls to outside servers or memory services, and the model itself never sees a memory tool, it just receives remembered facts as if it always knew them. Installation is two commands through the Hermes plugin system. The download is fairly large, around 135 megabytes, because it bundles everything needed to run fully offline right away, including prebuilt libraries for Mac, Linux, and Windows and a small local model used to recall facts. Configuration is optional, with sensible defaults for how selective and how stable the memory behaves. This is a good fit for anyone using Hermes who wants an assistant, coding helper, or operations tool that keeps context between sessions without needing to rely on an external memory service or repeat instructions every time.
A local, offline memory plugin that lets the Hermes Agent remember facts across chat sessions without sending data to any outside service.
Mainly Python. The stack also includes Python, SQLite, ctypes.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.