Cohestry
A local-first second-brain tool designed for AI agents. It supports local storage of Markdown notes, automatic linking, a 3D knowledge graph, and retrieval through an MCP bridge.
Timeline (last 90 days)
Cohestry is a "second brain" tool designed specifically for AI Agents, whose core features include local Markdown storage, automatic note linking, a 3D knowledge graph and MCP-bridged retrieval.
All of Cohestry's Markdown notes are stored locally and do not pass through the cloud.
The Cohestry system identifies related notes and automatically establishes connections.
Cohestry provides a 3D knowledge graph, and as usage time grows, knowledge accumulates into a visualized network.
Through the MCP bridge, Cohestry lets Agents actively search the memory store and retrieve notes on demand, rather than loading an entire folder into the context window.
Before handing content to the Agent, Cohestry desensitizes confidential information such as API keys and p8 keys.
In Cohestry, the memory store exposes tools such as "search notes" and "read notes" in the form of an MCP Server, and the Agent simply initiates a tool call when needed.
Agent long-term memory products currently fall roughly into two approaches: hosted cloud memory services, and solutions like Cohestry that emphasize local-first and user control.
Cohestry uses human-readable Markdown as the underlying format for memory.
What Cohestry currently offers is more a personal usage experience; data from scale use is still lacking to test automatic association quality, graph usability, and MCP retrieval recall precision.
All Facts (10)
Cohestry is a "second brain" tool designed specifically for AI Agents, whose core features include local Markdown storage, automatic note linking, a 3D knowledge graph and MCP-bridged retrieval.
50%UnverifiedAll of Cohestry's Markdown notes are stored locally and do not pass through the cloud.
50%UnverifiedThe Cohestry system identifies related notes and automatically establishes connections.
50%UnverifiedCohestry provides a 3D knowledge graph, and as usage time grows, knowledge accumulates into a visualized network.
50%UnverifiedThrough the MCP bridge, Cohestry lets Agents actively search the memory store and retrieve notes on demand, rather than loading an entire folder into the context window.
50%UnverifiedBefore handing content to the Agent, Cohestry desensitizes confidential information such as API keys and p8 keys.
50%UnverifiedIn Cohestry, the memory store exposes tools such as "search notes" and "read notes" in the form of an MCP Server, and the Agent simply initiates a tool call when needed.
50%UnverifiedAgent long-term memory products currently fall roughly into two approaches: hosted cloud memory services, and solutions like Cohestry that emphasize local-first and user control.
50%UnverifiedCohestry uses human-readable Markdown as the underlying format for memory.
50%UnverifiedWhat Cohestry currently offers is more a personal usage experience; data from scale use is still lacking to test automatic association quality, graph usability, and MCP retrieval recall precision.
50%