Store and retrieve AI memories with configurable decay and reinforcement across short-term JSONL and long-term Markdown vaults. For developers building AI agents that need persistent, time-aware context management.
io.github.simplemindedbot/mnemex
Local install
STDIO
No auth required
How models use it and what it is built for.
Store and retrieve AI memories with configurable decay and reinforcement across short-term JSONL and long-term Markdown vaults. For developers building AI agents that need persistent, time-aware context management.
Local install — runs as a subprocess.
Configuration this server reads at startup.
Storage path for short-term memory JSONL files
Path to Obsidian vault for long-term memory
Decay model: power_law (default), exponential, or two_component
Half-life in days for power-law decay model
Where to find authoritative docs and source for Simplemindedbot Mnemex.
Paste any of these into Agent Studio after connecting Simplemindedbot Mnemex.
Common questions about connecting and running Simplemindedbot Mnemex.
What is the difference between short-term and long-term memory in Mnemex?
Short-term memory uses JSONL files (configured via MNEMEX_STORAGE_PATH) with time-based decay, while long-term memory persists in an Obsidian vault (LTM_VAULT_PATH) as Markdown. Memories decay over time in short-term storage based on your chosen decay model.
How do I configure memory decay for my AI agent?
Set MNEMEX_DECAY_MODEL to power_law (default), exponential, or two_component, then adjust MNEMEX_PL_HALFLIFE_DAYS to control how quickly memories fade. Power-law decay is the default and uses a configurable half-life in days.
Do I need an Obsidian vault to use Mnemex?
No—short-term memory works with just MNEMEX_STORAGE_PATH. Long-term memory requires an Obsidian vault path (LTM_VAULT_PATH) only if you want persistent, non-decaying storage alongside short-term memories.
What decay models does Mnemex support and when should I use each?
Mnemex offers power_law (realistic forgetting curve, default), exponential (faster decay), and two_component (hybrid). Power-law is recommended for human-like memory behavior; choose exponential for rapid forgetting or two_component for mixed retention patterns.
How do I install and run Mnemex?
Install via uvx mnemex@0.5.2, then configure MNEMEX_STORAGE_PATH and optionally LTM_VAULT_PATH. The server runs over stdio and integrates with any MCP-compatible AI client.
MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 60+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.
Open Agent StudioJudges Panel
45 judges that evaluate AI-generated code for security, cost, and quality with built-in AST.
Addozhang Nexus
Query Sonatype Nexus Repository (OSS/Pro) for Maven, Python, and Docker artifacts
Cyanheads MCP Ts Template
TypeScript template for building MCP servers with declarative tooling, observability, and auth.
Chernistry Bernstein
Declarative agent orchestration for engineering teams