Pi Hermes Memory plugin
Memory for the Pi agent: facts and lessons persist, past chats are searched via SQLite and API keys never reach memory.
- 529 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The extension stores conversation content on disk and saves conclusions automatically.
Install
Manual install
pi install npm:pi-hermes-memoryThe first README Quick Start command. Then `/memory-index-sessions` indexes past sessions.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The extension stores conversation content on disk and saves conclusions automatically.
This is third-party code. Review the repository files before installing.
What it does
The extension gives Pi persistent memory and session search. Facts, preferences and corrections are stored in Markdown files, while a SQLite FTS5 mirror lets you search every past conversation: what did we decide about auth. The agent remembers failures with their reasons, sorts entries into categories (failures, corrections, insights, conventions, tool quirks) and saves solutions as procedural skills. Background learning runs every ten turns, and when memory fills, entries merge without loss. A secret scanner blocks keys and tokens from being saved.
Who it is for. Pi users tired of explaining the same project rules to the agent in every new session.
Good fit when
- The agent steps on an error already analysed last week.
- You must find what was discussed on a topic a month ago.
- Project conventions should be picked up without reminders.
- Solving a hard task should be saved as a reusable instruction.
Not a fit when
- The agent is not Pi: the extension is built for it alone.
- Conversation history may not be stored locally.
- Team-wide shared memory is needed: storage sits on the user's machine.
Example request
Search memory for what we decided about auth and remind me which approaches failed.Limitations
The extension works only in Pi. Background learning saves conclusions from conversations, so your working details end up in memory too, and secret protection does not catch every possible key form. Storage sits in `~/.pi/agent/pi-hermes-memory`, and access to that directory should be restricted. Data migrates automatically from older versions, so launch Pi once after an upgrade and check the result.
How to disable. Remove the package from Pi and delete `~/.pi/agent/pi-hermes-memory` with the memory entries and SQLite index.
FAQ
How do I install it?
With `pi install npm:pi-hermes-memory`.
How do I index past conversations?
With `/memory-index-sessions`, once is enough.
What does `/memory-sync-markdown` do?
Moves older Markdown entries into SQLite search; the step is optional.
How do I learn what it can do?
The `/learn-memory-tool` command explains how the tool works.
What about secrets?
API keys and tokens are blocked at save time, but entries are still worth reviewing.
Discussion
No questions or reviews yet.
Related
PluginsLocal memory for agents with a review queue: SQLite, an audit log and a token ledger, one vault for Claude Code, Codex and Cursor
MCP server with a memory of the agent's own: it decides what to keep and wakes with it each session, stored in one SQLite file
External memory for Hermes, Claude Code, Codex and Cursor: session archive, recall layers and relevant context injected back
Coding memory across sessions, machines and teammates: decisions and dead ends flow into later prompts on their own.
Local persistent memory for coding agents on SQLite: decisions, fixes and code context survive between sessions.
Long-term memory for agents: a remote MCP server, plugins for Claude Code, Cursor and Codex, and an API for your own apps
README in short
The README lists memory features, a quick start with Pi commands, upgrade notes for 0.7.10 and a feature table.