M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
P

PMB AI

Community
Local-first persistent memory for AI coding agents.
How to install ↓ Compare GitHub source repository ↗
Category
Dev Tools #74 of 438
Stars
★ 285 Very popular
Transport
stdio (local process)
Runtime
Python
Credentials
No credential needed
License
Apache-2.0
Last commit
Tools
13

Static review · not runListed 2026-08-07

FMRS 63/100 · C · Weakest Reliability 10/20 — Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesConnects to a database

Best for

  • Developers using Claude Code, Cursor, or Codex
  • Users prioritizing privacy and local data control
  • Teams needing persistent memory without cloud dependency

Not for

  • Users requiring online collaboration or cloud sync
  • Users without local disk space or needing real-time sync across devices
  • Users wanting zero configuration and avoiding command-line operations

Required permissions

  • Read local filesystem (for indexing project, PDFs, etc.)
  • Write to local SQLite database and LanceDB indexes
  • Execute CLI commands (pmb)
  • Optional network access (e.g., when using Ollama or OpenAI for summarization)

Risks and side effects

  • Local data loss risk if not backed up
  • Indexing large projects may consume disk space
  • Data is not encrypted by default; secrets are auto-redacted
  • Requires command-line familiarity

Setup

Before you start

Runtime：Python

  1. Install with pip install pmb-ai.
  2. Run pmb setup to detect your agent and wire the MCP entry.
  3. Run pmb warmup to preload the model (first recall is instant).
  4. Restart your agent, then just talk - memory is automatic.
  5. Use pmb stats to see what's stored, pmb recall "query" to search, and pmb doctor to confirm everything is wired.

Check that it works

After install, run pmb doctor to confirm wiring, restart your client, and call the prepare(message) tool — it should return project_context, lessons, and open_goals in a few milliseconds.

Troubleshooting

  1. Common issues:
  2. - Cold start slow: run `pmb warmup` to preload model.
  3. - Agent not using memory: check `pmb connect` setup, or use `pmb hooks install` to force injection.
  4. - Slow writes: ensure `mcp.record_batch_async` is true.
  5. - Multi-agent conflicts: SQLite WAL handles concurrent writes, but be mindful of locks.
  6. - Indexing failures: check file permissions and paths.

Things to try

Once connected, you can ask your AI assistant things like:

  • Call prepare with my message and show me the key decisions and open goals for this project
  • Use recall to find the auth decision we made earlier
  • Record a decision: we standardize on PostgreSQL 16 as the database
  • What was the historical value of the user.city keyed fact?

Tools 13

prepare read-only
Injects matching memory context (project overview, lessons, decisions, recent activity) before the agent thinks.
recall read-only
Hybrid search over memory using BM25, vector, entity graph, and optional reranking.
record_keyed_fact writes
Stores personal facts (e.g., user city) that can change; old values are archived, never lost.
record_batch writes
Batch records events asynchronously with sub-millisecond return.
record_decision writes
Records a decision made by the agent.
record_lesson writes
Records a lesson as a rule to follow later.
record_goal writes
Records a goal to track progress.
index writes
Indexes project code, PDFs, or directories, extracting structure and content.
Show 5 more tools
track writes
Tracks modules or changes, generating summaries (e.g., commit intent).
import writes
Imports history from external sources (e.g., ChatGPT export).
forget destructive
Archives or deletes a memory.
mark_lesson_followed writes
Marks whether a lesson was followed, enabling the self-improvement loop.
keyed_fact_as_of read-only
View historical value of a keyed fact at a specific time.

Use cases

Preserve agent memory across sessions, avoiding re-explaining decisions and constraints.
Store project structure, PDFs, personal facts, and other long-term information.
Share memory across a team via optional HTTP mode with authentication.
Enhance agent behavior through lesson rules and follow-through feedback.

Supported clients

Claude Desktop
Cursor
Codex
Windsurf
Zed
VS Code

Listed from the project's documentation, not tested by this site.

Overview

PMB gives Claude Code, Cursor, Codex and other MCP-aware agents a real memory: decisions you made last week, lessons you taught them, personal facts, project structure, PDFs. They survive every restart, every model upgrade, every agent switch - because they live in a local workspace you own, with SQLite as the durable source of truth and rebuildable search indexes beside it. No cloud, no API keys, no LLM call on the read path. Just local files.

Similar servers

Context7 80 · B

Upstash's official server providing up-to-date third-party library docs for AI coding assistants

★ 63.3k · Tools 2 Compare with this →
Compare with other MCP servers

Source revision 59b6800cfed6 Data synced 2026-10-19 Read the FMRS scoring method