Perseus

•
43
•

Provides live, verified workspace context to AI agents, resolving git status, running services, and test results into a ready-to-use format.

Perseus functions as a live context engine and Model Context Protocol (MCP) server designed to eliminate the 'cold start' problem for AI assistants. Instead of agents spending turns orienting themselves, Perseus pre-processes directives within source documents, resolving dynamic information like active services, recent test outcomes, and project checkpoints. It then outputs verified, plain markdown, ensuring that AI assistants receive up-to-date, factual context, significantly reducing token usage and improving efficiency for development and agent-driven workflows.

Key Features

01Integrates with major AI assistants (Claude, Cursor, Hermes, Codex, Rovo Dev)
02Resolves live workspace context from git, services, and tests
03Implements Model Context Protocol (MCP) for tool exposure to AI assistants
04Includes Mnēmē in-process persistent memory and Mimir for advanced memory management
0520 GitHub stars
06Offers performance optimizations for token reduction and cold-start elimination

Use Cases

01Integrating dynamic, real-time project context into AI agent workflows
02Automating the setup and orientation process for AI assistant sessions
03Enabling persistent, cross-session memory and knowledge recall for AI agents