Sage
Give coding agents persistent, consensus-validated institutional memory that runs on your machine instead of losing context every session.
Overview
io.github.l33tdawg/sage is a MCP server for the Build phase that provides locally run, consensus-validated institutional memory for AI agents.
What is this MCP server?
- Persistent institutional memory stored and queried locally for privacy-sensitive solo builds
- Consensus-validated writes so contradictory agent claims do not pollute long-term memory
- OCI image ghcr.io/l33tdawg/sage:4.0.0 with stdio transport for MCP clients
- Version 4.0.0 indicates a mature local-memory server aimed at production agent stacks
- Runs locally—no mandatory cloud memory vendor for indie workflows
- server.json version 4.0.0
- OCI identifier ghcr.io/l33tdawg/sage:4.0.0
Community signal: 229 GitHub stars.
What problem does it solve?
Agents forget prior decisions and pollute memory with conflicting statements unless persistence and validation live behind MCP tools you control.
Who is it for?
Privacy-conscious solo builders who want durable agent memory without shipping conversation logs to a third-party memory API.
Skip if: Builders who only need ephemeral chat context or a full team wiki with human-only editorial workflows.
What do I get? / Deliverables
After install, agents read and write institutional memory on your machine with consensus validation exposed through stdio MCP.
- stdio MCP access to persistent institutional memory tools
- Consensus-validated memory layer for multi-session agent work
- Local-only memory path suitable for sensitive indie projects
Recommended MCP Servers
Journey fit
Agent-tooling on Build is where you attach memory and validation layers that make long-horizon agent work possible. Institutional memory for agents is core agent-tooling: it is the MCP shelf for local persistence and consensus checks before you ship workflows that depend on remembered facts.
How it compares
Local institutional-memory MCP server, not a cloud note app integration or a generic RAG skills pack.
Common Questions / FAQ
Who is io.github.l33tdawg/sage for?
Developers running local coding agents who need memory that survives sessions and resists contradictory agent updates.
When should I use io.github.l33tdawg/sage?
Use it while building agent workflows where architecture, customer facts, or runbooks must persist and be consensus-checked across many agent turns.
How do I add io.github.l33tdawg/sage to my agent?
Run the Sage OCI container ghcr.io/l33tdawg/sage:4.0.0 with stdio wired into your MCP client config, then enable the memory tools in your agent host.