Back to Apps

ProjectMem
by riponcm
Local-first memory and judgment layer for AI coding agents. Stop repeating failed attempts and save tokens with durable project context.
0 stars
Works in:CursorCodex
Exposes:Tools
What it does
ProjectMem provides a local-first memory and judgment layer that sits above AI coding tools. It captures failed debugging attempts, architectural decisions, and project-specific 'gotchas', injecting this experience back into future sessions so AI agents start experienced rather than amnesiac.
Tools
get_summary(): Provides a distilled overview of project memory.precheck_file(path): Surfaces known failure history for a specific file before editing.log_issue(summary, location): Records a new bug or problem encountered.record_attempt(summary, outcome): Logs a fix attempt and its result (failed/partial/worked).add_decision(summary, supersedes): Records architectural decisions and can retire stale ones.get_score(): Outputs a prevention score (A+ to F) and ROI metrics.get_global_gotchas(library): Inherits lessons learned from other projects using the same library.
Installation
"mcpServers": {
"projectmem": {
"command": "python",
"args": [
"-m",
"projectmem.mcp_server",
"--root",
"/absolute/path/to/your/project"
]
}
}
Supported hosts
Confirmed support for Claude Desktop, Cursor, Antigravity, and Codex.
Quick install
pip install projectmem && pjm initInformation
- Pricing
- free
- Published
- 6/30/2026
- stars
- 0
Categories
Choose your AI client and follow the steps below.
Codex
Add to ~/.codex/config.toml using [mcp_servers.projectmem] section.Cursor
Add to mcp.json in ~/.cursor/ or project .cursor/ folder.Claude Desktop
Add to claude_desktop_config.json using the python -m projectmem.mcp_server command with --root flag.





