A memory layer for coding agents that stores and retrieves repo context, conventions, and invariants.
gps is easy to set up with strong trust signals. Check agent compatibility and use-case fit before adding it to your workflow.
gh repo view invariance-ai/gps --webOpen the official repository or website.
Check the README for package manager, auth, and platform requirements.
Try it in a small test task inside your agent workflow.
GPS is a tool that helps AI coding assistants remember important details about your codebase. You teach it once—like your project's rules, preferences, and past decisions—and it automatically shows the relevant info when you or your agent work on related code. It works with Claude Code, Codex, Cursor, and other coding agents.
gps is a memory layer for coding agents such as Claude Code, Codex, Cursor, and shell-based agents. It stores repo conventions, best practices, user preferences, invariants, decisions, TODOs, test commands, and lessons from past edits. When an agent is about to edit code, gps retrieves the relevant slice of memory to guide the agent. It integrates via MCP (Model Context Protocol) and provides a CLI for setup, capturing notes, preparing edits, and post-edit auditing. Key features include: persistent notes and invariants, automatic capture from session transcripts, decision tracking, and a core loop of 'gps next', 'gps prepare', 'gps remember', and 'gps done'. It pairs with tars (another tool) for learning orchestration methods. gps is open-source under MIT license and is designed to reduce repetitive instructions and improve agent consistency.
Strong trust signals; still review the README and permissions before production use.
Last commit was about 6 days ago.
18 GitHub stars indicate community interest.
0 open issues signal maintenance load.
MIT license detected.
Store and retrieve project-specific conventions (e.g., 'always use the shared errors module')
Capture hard-won lessons from past edits so agents don't repeat mistakes
Prepare decision-ready briefs before making changes to a symbol
Post-edit self-audit to check changed symbols, tests, and invariants
Integrate with Claude Code, Codex, or Cursor to automatically inject relevant context
gps stores codebase context locally; ensure sensitive data is not captured in notes or invariants.
MCP integrations allow external tools to read/write memory; review security policy before enabling.
Automatic capture from transcripts may inadvertently store private information; configure capture policy carefully.
18
Stars
3
Forks
0
Issues
MIT
License
A project template that turns your repository into stable infrastructure for AI-assisted development.
Local code intelligence MCP server and CLI for AI coding agents, providing semantic search and call graph analysis.
Up-to-date code documentation for LLMs and AI code editors, eliminating outdated or hallucinated API references.
3 security/trust notes recorded.
Setup difficulty is 2/5.