PMB
Local-first persistent memory for AI coding agents via MCP.
PMB is a practical, private alternative to cloud memory services. It delivers fast hybrid recall on your own disk with no per-query cost, and its honesty scoring keeps memory useful. The CLI setup and lack of native cloud sync mean it's best for developers who want control and are okay with configuration.
Verified 4d ago · liveness 69/100 · cite: rightaichoice.com/tools/pmb
- Developers using AI coding agents who want persistent context across sessions
- Teams working with multiple agents who need shared memory on the same machine
- Users who prioritize data privacy and offline-first tools
- Developers tired of re-explaining project context each session
- Users who need cloud sync or team sharing natively
- Non-technical users unfamiliar with CLI or MCP setup
- Those seeking a turnkey product with no local configuration
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip PMB if you need cloud sync, team collaboration, or a zero-configuration setup, or if you're not comfortable with CLI and MCP integration.
No hidden costs—PMB is free and open source, but you'll spend time on initial setup and configuration.
PMB is free (Apache 2.0), making it ideal for developers and small teams who want a no-cost, self-hosted memory solution. Compared to cloud memory services that charge per query or subscription, PMB has no per-query cost, but it requires local setup and lacks native cloud sync.
In short
PMB — Local-first persistent memory for AI coding agents via MCP. Best for Developers using AI coding agents who want persistent context across sessions, Teams working with multiple agents who need shared memory on the same machine, Users who prioritize data privacy and offline-first tools. Free to use.
What people actually say about PMB — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
81 mentions across 6 sources (Hacker News, Product Hunt, App Store, Bluesky, GitHub, Lemmy) · researched Jul 4, 2026.
- +Local-first: all data stays on your disk, no cloud or API keys.
- +Hybrid recall (BM25 + vectors + graph) delivers relevant memory in ~35 ms.
- +Automatic write and read hooks integrate with Claude Code, Cursor, Codex, Zed.
- +Honest impact scoring flags dead memories that agents don't follow.
- +Open source under Apache 2.0, inspectable and exportable.
- −No cloud or team-synced version, only local storage.
- −Stale or reversed decisions can still surface without manual pruning.
- −Automatic memory injection uses context window tokens, reducing solution runway.
- −Only one maintainer; open issues may take time to resolve.
- −Requires manual pruning to remove irrelevant or dead memories.
- • No hidden costs; all features are free and open source.
Viability Score
How well maintained and how widely used is PMB? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: August 2026
How we score →Key Features
- Persistent SQLite memory file on disk
- MCP-native integration with Claude Code, Cursor, Codex, Zed
- Sub-millisecond classification and recall (~35 ms)
- Auto-inject relevant lessons, decisions, and project overview on every prompt
- Hybrid recall: BM25 + dense vectors + entity graph + RRF
- Honest impact scoring: tracks whether each lesson is followed, flags dead memories
- Local web dashboard with interactive entity graph (Map) and git-style timeline
- Async writes: SQLite first, embedding and vector insert on background thread
- Works offline, no API keys, no telemetry
- Open source under Apache 2.0 license
- Inspectable and exportable memory chunks
- Entity nodes color-coded by type, sized by importance
- Multi-project support with navigation lanes
- Optional local LLM integration (Ollama) for summarization and graph extraction
- Benchmarks: 94.6% recall@10 on LoCoMo, 88ms p50 recall at 2k memories
About PMB
PMB is a local-first, open-source memory system that gives AI coding agents persistent context, right on your disk. Instead of re-explaining your project, conventions, and past bugs every session, PMB stores decisions, lessons, and project facts in a single SQLite file. It automatically injects relevant memories into the agent's context before it reasons, using hybrid recall that fuses BM25, dense vectors, and an entity graph. No LLM call on the read path, so retrieval is fast—around 35ms—and fully offline.
Behind the Verdict
PMB stands out for its local-first, privacy-preserving approach. It stores memory in SQLite and vectors in LanceDB, all on your disk, with no cloud dependency. This means your data never leaves your machine, and you can inspect and export everything. For developers tired of re-explaining context to agents like Claude Code, Cursor, Codex, or Zed, PMB offers a single shared memory across all of them. The auto-recall on every prompt is sub-millisecond, and the write path is async so it never blocks your workflow. The honesty scoring is a unique feature: it tracks whether each lesson is actually followed, flagging dead memories so your context stays lean. However, PMB requires CLI setup and MCP integration, which might be a barrier for non-technical users. There's no built-in cloud sync or team collaboration, but for individual developers or small teams comfortable with local tools, this is a solid choice. If you prioritize privacy and control and are okay with a bit of terminal work, PMB is worth trying. Compared to cloud-based memory services, it's free and fully self-hosted, but you sacrifice convenience and shared access.
Researching PMB? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas PMB actually fits — and what changes day-one when you adopt it.
You're working on a project and tired of re-explaining your architecture and conventions every session.
Outcome: Install PMB, connect Claude Code, and it auto-recalls lessons and decisions, saving you time and keeping context consistent.
You want to share context between two different AI coding agents without manual copying.
Outcome: With PMB, both agents read the same memory, so you can switch tools seamlessly and maintain continuity.
Use Cases
- Persist project conventions and decisions so Claude Code remembers them across sessions
- Share context between Cursor and Codex without manual copying
- Track which coding rules your agent actually follows and prune ineffective ones
- Visualize your project's memory as an entity graph to spot knowledge gaps
- Review a git-style timeline of decisions and lessons learned over the project lifecycle
Models Under the Hood
as of 2026-08-27
Limitations
- This tool is local-first, storing memory in a single SQLite file on your disk.
- It is not a cloud service and requires no API keys or telemetry.
- Setup involves installing via pip and integrating with MCP-compatible tools like Claude Code, Cursor, Codex, and Zed.
- Memory is persisted locally, and there is no mention of multi-user collaboration or cloud sync.
as of 2026-08-21
Verification history
We have re-verified PMB 8 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
Showing the 6 most recent of 8 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published PMB tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Open Source
$0
Ideal for
Solo developers and teams who want a free, self-hosted memory solution for AI coding agents, comfortable with CLI setup.
What this tier adds
This is the only tier—free, open source under Apache 2.0, with all features included.
Where the pricing makes sense
The company stage and team size where PMB's pricing actually pencils out — and where peers do it cheaper.
PMB is free (Apache 2.0), making it ideal for developers and small teams who want a no-cost, self-hosted memory solution. Compared to cloud memory services that charge per query or subscription, PMB has no per-query cost, but it requires local setup and lacks native cloud sync.
Setup time & first value
How long it actually takes to get something useful out of PMB — broken out by persona, not the marketing-page minute.
For a developer familiar with CLI, setup takes about 60 seconds: pip install pmb-ai, run pmb connect <agent>, and you're ready. Non-technical users may need extra time to understand MCP concepts, but the dashboard helps visualize memory.
Switching to or from PMB
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From built-in agent memory: PMB stores memory in SQLite, so you can export existing notes or lessons and import them as memory chunks.
- ↗To cloud-based memory service: Export your SQLite file and import into a cloud service, though you'll lose local-first benefits.
Integrations
Resources & Guides
Tutorials & Learning
Official links
Tools that pair well with PMB
Common stack mates teams adopt alongside PMB, with the specific reason each pairing earns its keep.
Featured Head-to-Head Comparisons
Pmb vs Spider Cloud
For developers who need persistent context for coding agents and value data privacy, PMB is a free, offline-first solution. For teams building AI agents that require real-time web data for RAG, Spider Cloud offers a scalable, cost-efficient scraping API with advanced anti-detection. Choose PMB if your pain is repetitive context loss; choose Spider Cloud if you need structured data from the web.
Pmb vs Voyage Ai
Voyage AI is built for enterprises needing high-accuracy, domain-specific retrieval at scale, while PMB targets developers who want simple, private, local memory for coding agents. Choose Voyage if you run a production RAG system on sensitive data; choose PMB if you're tired of re-explaining context to Claude Code or Cursor.
Pmb vs Temporal Ai
If you need to orchestrate mission-critical AI agents or microservices with automatic retries and crash recovery, Temporal is the clear choice — its durable execution platform is battle-tested at scale. For developers who just want their coding agent (Claude Code, Cursor) to remember project context without cloud dependencies, PMB offers a lightweight, free, and privacy-first solution. Choose Temporal for production workflows; choose PMB for a smarter coding assistant.
Alternatives to PMB
View allPipedream
Developer-first platform to connect APIs, AI, and databases via workflows and agents.
Imcodes
Self-hosted shared memory, MCP tools, and cross-agent audit for multi-model coding workflows.
Hermes Webui
Self-hosted AI agent with persistent memory, cron scheduling, and multi-surface access.
Frequently Asked Questions
Used PMB? Help shape our editorial sentiment research.