What is project-memory-mcp AGENTS.md?

A set of instructions for a local MCP server that gives coding agents persistent memory across project folders. It stores project details, decisions, and development issues in plain files on disk.

Use it when working across multiple projects and recording or searching project notes and bugs. The server communicates with MCP clients such as Claude Code and Cursor.

It lets agents reuse selected project knowledge between sessions without depending on temporary conversation context.

project-memory-mcp AGENTS.md — Instructions file for Codex and OpenCode by kaaustubh: A set of instructions for a local MCP server…

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add instructions/kaaustubh/project-memory-mcp/agents-md

Clone the repo

git clone --depth 1 https://github.com/kaaustubh/project-memory-mcp

Made for: Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/kaaustubh/project-memory-mcp/agents-md.svg)](https://agentmods.dev/instructions/kaaustubh/project-memory-mcp/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/kaaustubh/project-memory-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/kaaustubh/project-memory-mcp/agents-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 7,988 This file is loaded in full into every session.

When invoked 7,988 The same file — it is already loaded in full.

Security scan B 1 finding. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.07988	$0.07988
Opus 5	$0.03994	$0.03994
Sonnet 5	$0.01598	$0.01598
Haiku 4.5	$0.00799	$0.00799

Measured today against content hash bd7c4a588b39, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

project-memory-mcp AGENTS.md scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- 2026-06-05 (v1.3.0): Shipped the previously-deferred guaranteed-capture Stop hook, but as a SEPARATE concern from the MCP server: it's a `hook` subcommand of index.js, registered via `install-hook` into `~/.claude/sett

AGENTS.md · 122 lines

How it starts

The opening of the file, as written. The whole thing — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.

project-memory-mcp — Agent Memory

This is the tool's own dev memory. It is auto-loaded when you open this repo in an editor, but it is NOT shipped to npm (the files whitelist excludes it). Note: the server excludes itself from project discovery (.memory-server is dotfile- and SKIP-filtered in index.js), so this project won't appear in list_projects or an all-projects search_issues — it's first-class only via the editor auto-load here.

What this is

A small local MCP server that gives AI coding agents (Claude Code, Cursor, any MCP client) a shared, persistent memory of a folder of projects: what each project is, key decisions, and every bug/issue faced during development.

npm: @kaaustubh/project-memory-mcp · repo: github.com/kaaustubh/project-memory-mcp
All logic lives in a single file: index.js (~210 lines).

Architecture (the core ideas)

Protocol: MCP = JSON-RPC 2.0 over stdio. The client spawns this script as a subprocess and exchanges newline-delimited JSON messages (initialize → notifications/initialized → tools/list → tools/call).
Stateless: every tool reads/writes plain files on disk and returns; nothing is held in memory between calls. Consequence: multiple clients (and machines) sharing the same files share one memory. The filesystem is the source of truth.
Two-layer memory (push vs pull):
- Push — <project>/AGENTS.md is auto-loaded into the model's context by the editor (via CLAUDE.md → @AGENTS.md), NOT by this server. Low-volume, durable: identity, decisions, learnings. Keep lean.
- Pull — <project>/issues.jsonl is NOT auto-loaded; high-volume bug history fetched on demand via search_issues. Keeps always-loaded context small but searchable.
Discovery is the filesystem, not a registry (listProjectDirs): scan ROOT for subdirs containing AGENTS.md. Can't drift.

Stack & layout

Node.js ESM ("type": "module"), Node >= 18.
Deps: @modelcontextprotocol/sdk (high-level McpServer + StdioServerTransport), zod (tool input schemas).
index.js — server + all 16 tools + the install subcommand (gated on argv[2]).
install.sh — install path for the from-source (git clone) install.
README.md — public docs (this is what shows on npmjs.com).
ROOT = PROJECT_MEMORY_ROOT env, else parent dir of the script.

Read the full file on GitHub · 122 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

today Changed · +3 lines · +557 tokens per session bd7c4a588b39
4d ago First seen · 119 lines · 7,431 tokens per session scan B 8dabe21f6be4

Subscribe to this mod's changes

project-memory-mcp AGENTS.md is an instructions file published in the GitHub repository kaaustubh/project-memory-mcp (1 stars, last pushed yesterday), licensed MIT. It adds 7,988 tokens to every session, about $0.0399 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

LLM Memory and context ai-agents claude-code cursor developer-tools mcp

lean-ctx AGENTS.md

AGENTS.md instructions for yvgude/lean-ctx, covering context engineering layer, mandatory multi-agent delivery, integration mode: replace, subagent compatibility (cursor / claude code) and cli commands (optimized shell, lower overhead).

yvgude/lean-ctx · 2,223 tokens

lean-ctx copilot-instructions.md

Copilot instructions for yvgude/lean-ctx, a project described as: Control what your AI can see. LeanCTX (Lean Context) is the context intelligence layer for AI agents — one local Rust binary that decides what they read, remembers what they learn, guards what they touch, and proves what they save. 60–90% fewer tokens…

yvgude/lean-ctx · 253 tokens

project-memory-mcp AGENTS.md