Plur is live · free · last verified 23 Oct 2026 · viability monitor
Plur
Open YAML agent memory: one shared engram store for Claude Code, Cursor, Windsurf, and Codex
One specific fact, check it yourself: PLUR publishes a 97.6% R@5 on LongMemEval with fully local retrieval, and the memory files are YAML you can open. If MCP is your stack and you run several agents, this is the most inspectable cross-agent memory fix around. Hosted multi-user platforms are a different purchase.
Verdict generatedVerified 7d ago · liveness 64/100 · cite: rightaichoice.com/tools/plur
- Engineers running Claude Code, Cursor, Windsurf, or Codex side by side who want one shared memory store
- Privacy-conscious teams that need agent memory to stay on local disk rather than a hosted service
- Developers who want to read, edit, or delete what an agent remembers as plain YAML
- Teams distributing reusable agent knowledge through Knowledge Packs
- Non-technical users who need a GUI, dashboard, or managed onboarding
- Anyone running a single AI assistant where a cross-tool store adds nothing
- Workflows built on assistants that do not speak MCP
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 PLUR if your agents don't speak MCP or you want a hosted, dashboard-managed memory service — PLUR's value comes from a shared local store you inspect and edit as YAML yourselves.
Someone on your team has to own the ~/.plur/ store — correcting wrong engrams, resolving pack conflicts, and pruning on a schedule is unpaid maintenance work.
PLUR's memory engine ships as open-source Apache-2.0 software with the engram spec published freely, so the direct cost of adoption is engineering time, not a subscription. That is a different bargain from hosted agent-memory platforms such as Mem0, Letta, or Zep, where you pay per seat or per request for infrastructure you don't operate. For a team already running several MCP-speaking coding agents, the comparison is PLUR plus your own maintenance versus a hosted bill plus none of the
In short
Plur — Open YAML agent memory: one shared engram store for Claude Code, Cursor, Windsurf, and Codex. Best for Engineers running Claude Code, Cursor, Windsurf, or Codex side by side who want one shared memory store, Privacy-conscious teams that need agent memory to stay on local disk rather than a hosted service, Developers who want to read, edit, or delete what an agent remembers as plain YAML. Free to use.
What is Plur?
Plur — Open YAML agent memory: one shared engram store for Claude Code, Cursor, Windsurf, and Codex. Best for Engineers running Claude Code, Cursor, Windsurf, or Codex side by side who want one shared memory store, Privacy-conscious teams that need agent memory to stay on local disk rather than a hosted service, Developers who want to read, edit, or delete what an agent remembers as plain YAML. Free to use.
PLUR is an open, local-first memory layer for AI coding agents. You write knowledge once to a shared store of engrams — typed units of learned knowledge like a correction, a preference, or a project fact — kept as plain YAML on your own disk. The MCP server exposes that store to any MCP-compatible host, so a correction made in Claude Code is already there the next time Cursor, Windsurf, or Codex runs.
Key features
- Local-first engram memory stored as plain YAML in ~/.plur/
- Human-readable memory files you can open, edit, or delete directly
- Corrections apply on the next turn — no retraining or model update
- One shared store read by Claude Code, Cursor, Windsurf, Codex, OpenClaw, and Hermes
- MCP tools: plur_learn, plur_inject_hybrid, plur_recall_hybrid, plur_capture, plur_ingest, plur_status
- Hybrid retrieval with BM25 plus embeddings and a reranker, fully local at zero API cost
- ACT-R decay fades engrams that stop being recalled
- Similarity search deduplication catches duplicates before an engram is written
- Structural fix recommendations when the same correction keeps recurring
- Knowledge Packs export curated engrams as installable YAML bundles
- Privacy scan blocks secrets and personal data on pack export
- SHA256 integrity hash per pack plus conflict detection on install
- Multi-store scopes (personal, project, team) read in one inject call
- Readonly team store sharing via plur stores add --scope --readonly
- MCP server for any MCP-compatible agent or custom framework
Behind the verdict
PLUR's pitch is narrow and unusually verifiable: YAML engrams on your own disk, retrieval with zero API spend, and a benchmark report you can read rather than a marketing number.
Models under the hood
as of 2026-10-25
Limitations
- PLUR is a local-first agent memory layer whose value depends on the tools you use speaking MCP and on you consistently correcting them; silence means nothing is learned.
- Engrams are stored as human-readable YAML in ~/.plur/, so keeping memory accurate and deciding whether a stored fact is still true falls to the user.
- The engram specification is new, first published March 2026, so surrounding tooling and third-party implementations are still maturing.
- The primary interface is the CLI and the config directory rather than a dashboard.
as of 2026-10-13
Researching Plur? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is Plur still active?
Plur is live · free · last verified 23 Oct 2026 · viability monitor.
Viability score
How well maintained and how widely used is Plur? 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: October 2026
How we score →What's new in Plur
Checked 10 days agoAcross the latest 7 updates: 7 news mentions.
Test Agent Memory Corrections Without Losing Unrelated Preferences
PLUR publishes an evaluation checklist for changing one remembered rule and verifying unrelated agent preferences survive across fresh sessions.
Choosing Agent Memory: Can You Trace a Recalled Fact to Its Source?
PLUR outlines a source-traceability trial for agent memory tools: separate assertions from inferences and inspect the recalled record.
Stop Agent Memory Loss: A Cross-Session Verification Checklist
PLUR posts a checklist tracing a forgotten correction through storage, retrieval and use, then testing it in a fresh session.
Your Agent Forgot Again: A Cross-Session Memory Restart Checklist
PLUR publishes a diagnostic that tests saving, scope, retrieval and use separately to explain why agents lose project context after restart.
Choosing an Agent Memory Tool: Test Duplicate Writes Before You Commit
PLUR proposes a test for memory retries and duplicate records, warning against conflating transport safety with semantic deduplication.
Choosing an Agent Memory Tool: Test Corrections Against Delayed Writes
PLUR publishes a vendor-neutral exercise checking whether delayed memory writes can undo an approved correction across sessions.
Choosing an Agent Memory Tool: Test Your Exit Plan First
PLUR publishes a data-exit drill for agent memory tools: inspect an export, preserve scope and provenance, test a replacement store.
What do users say about Plur?
What people actually say about Plur — is it worth it?
Our community sentiment research written Jul 3, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about Plur— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 23 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified Plur 11 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.
- — changed limitations, pricing page
- — changed description, our verdict, pricing tiers
- — changed features, pricing tiers, summary, our verdict, description, integrations, pricing page
- — changed our verdict, who should skip it, description, who it suits, summary, features, setup time, pricing fit, hidden costs, use cases, limitations, switching in, switching out
- — changed pricing page
- — changed pricing tiers, summary, our verdict, features, who it suits, who should skip it, description
Showing the 6 most recent of 11 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does Plur cost in 2026?
Plur: Free to use. Pricing last verified 23 Oct 2026.
Plans compared
Every published Plur tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Open Source (Apache-2.0) | Free | — | Engineering teams and solo developers running Claude Code, Cursor, Windsurf, or Codex who are willing to install the MCP server and manage a local YAML store themselves.Adds: Starting tier: the full local engine — engram spec, PLUR core, MCP server, CLI, Python SDK, local hybrid retrieval, decay, deduplication, and Knowledge Packs — at no license cost under Apache-2.0. |
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.
Where the pricing makes sense
The company stage and team size where Plur's pricing actually pencils out — and where peers do it cheaper.
PLUR's memory engine ships as open-source Apache-2.0 software with the engram spec published freely, so the direct cost of adoption is engineering time, not a subscription. That is a different bargain from hosted agent-memory platforms such as Mem0, Letta, or Zep, where you pay per seat or per request for infrastructure you don't operate. For a team already running several MCP-speaking coding agents, the comparison is PLUR plus your own maintenance versus a hosted bill plus none of the
Has Plur's pricing changed?
No plan price for Plur has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick Plur?
Pick Plur if you are Engineers running Claude Code, Cursor, Windsurf, or Codex side by side who want one shared memory store, Privacy-conscious teams that need agent memory to stay on local disk rather than a hosted service, Developers who want to read, edit, or delete what an agent remembers as plain YAML, Teams distributing reusable agent knowledge through Knowledge Packs.
Look elsewhere if you are Non-technical users who need a GUI, dashboard, or managed onboarding, Anyone running a single AI assistant where a cross-tool store adds nothing, Workflows built on assistants that do not speak MCP, Teams that need a permanent knowledge base — ACT-R decay drops engrams that stop being recalled.
Skip it if Skip PLUR if your agents don't speak MCP or you want a hosted, dashboard-managed memory service — PLUR's value comes from a shared local store you inspect and edit as YAML yourselves.
GeneratedReal-world workflow fit
Concrete scenarios for the personas Plur actually fits — and what changes day-one when you adopt it.
You install @plur-ai/mcp, tell Claude Code once that the team uses pnpm not npm, and the correction is written as an engram under ~/.plur/. Cursor's next session reads the same store through plur_inject_hybrid and stops suggesting npm install.
Outcome: The correction sticks from the next message onward with no retraining, and you can open the YAML file to confirm exactly what was stored.
You record project conventions — tag formats, file placement, API quirks — as project-scoped engrams, then export the curated set as a Knowledge Pack. The privacy scan blocks secrets and personal data, and the pack carries a SHA256 hash before anyone installs it.
Outcome: New engineers install one pack and inherit the same house rules every agent follows, with conflict detection flagging anything that clashes with an existing engram.
A subagent learns a pattern during a run and writes it to the shared store. The orchestrator's next plur_inject_hybrid call reads it alongside personal and team scopes, with no copying or syncing between agents.
Outcome: Knowledge propagates across the agent graph without a message-passing layer, and plur_status shows what changed.
Use cases
- Correct your agent's coding style once and have it hold across Claude Code and Cursor
- Store project conventions such as tag formats and file placement so every agent follows them
- Record API quirks and debugging insights as engrams your whole team reads
- Keep preferred communication tones and design preferences across sessions
- Track tool routing preferences among 100+ available tools without re-querying
- Distribute curated team knowledge to other developers as installable Knowledge Packs
- Share one engram store between subagents and an orchestrator over MCP
- Rehearse an agent-memory restart or restore before a long-running workflow depends on it
Setup time & first value
How long it actually takes to get something useful out of Plur — broken out by persona, not the marketing-page minute.
Solo engineer on Claude Code or Cursor: about ten minutes, since it's one command to enable the MCP server plus a first correction. Team lead setting conventions: an hour or two to write the initial engram set and export a Knowledge Pack, plus a review pass for the privacy scan. Multi-agent builder wiring subagents and an orchestrator: half a day to decide which scopes each agent reads and to
Switching to or from Plur
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From hand-written CLAUDE.md or .cursorrules prompt files: convert each durable rule into an engram statement, then let plur_inject_hybrid select only the relevant ones per task.
- →From pasted context blocks in your agent prompts: move recurring decisions and API quirks into project-scoped engrams so they persist across sessions.
- →From a team wiki page of conventions: export the curated rules as a Knowledge Pack so teammates install them rather than retype them.
- ↗To a hosted memory platform: export engrams as YAML, then re-ingest each statement into the target service; the ~/.plur/ layout is the interchange format.
Integrations
Resources & guides
Tutorials & Learning
YouTube returned 6 videos for “Plur”, and we withheld 6: 6 could not be judged, because “Plur” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Plur.
Official links
What are the best Plur alternatives?
The closest alternatives to Plur we list are Projectmem, Agentmemory and Mengram. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Projectmem | Free | Free | — |
| Agentmemory | Free | Free | — |
| Mengram | Freemium | Pricing not published | — |
Alternatives to Plur
View all
Projectmem
Local-first coding agent memory that records issues, failed attempts and fixes inside your repo and feeds them back to Claude Code, Cursor or Codex at session
Agentmemory
Agentmemory gives coding agents shared, local memory across Claude Code, Cursor, Codex and 22 other agents.
Tools that pair well with Plur
Common stack mates teams adopt alongside Plur, with the specific reason each pairing earns its keep.
Projectmem
Local-first coding agent memory that records issues, failed attempts and fixes inside your repo and feeds them back to Claude Code, Cursor or Codex at session
Agentmemory
Agentmemory gives coding agents shared, local memory across Claude Code, Cursor, Codex and 22 other agents.
Mengram
Mengram is a self-hostable AI memory API that gives Claude, Cursor, Codex and ChatGPT one shared memory graph.
Plur vs its closest rival
Plur's closest head-to-head is with Spider Cloud — Plur and Spider Cloud serve entirely different needs: Plur is a free, open standard for cross-tool AI agent memory (YAML engrams) that eliminates cloud dependency, while Spider Cloud is a freemium web crawling/scraping API with a Rust engine and AI-powered extraction. If you need persistent, inspectable memory across multiple coding agents, choose Plur. If you need to feed real-time web data into RAG pipelines or LLM workflows, Spider Cloud is purpose-built for that. Verdict generated
Head-to-head comparisons
Plur vs Spider Cloud
Plur and Spider Cloud serve entirely different needs: Plur is a free, open standard for cross-tool AI agent memory (YAML engrams) that eliminates cloud dependency, while Spider Cloud is a freemium web crawling/scraping API with a Rust engine and AI-powered extraction. If you need persistent, inspectable memory across multiple coding agents, choose Plur. If you need to feed real-time web data into RAG pipelines or LLM workflows, Spider Cloud is purpose-built for that.
GeneratedPlur vs Temporal Ai
For teams needing reliable, crash-proof execution of multi-step workflows—especially AI agents that must survive failures—choose Temporal. For developers who want transparent, editable, and portable agent memory across multiple coding assistants without cloud dependency, choose Plur. They solve orthogonal problems; the best pick depends on whether your pain point is durability or memory portability.
GeneratedPlur vs Presto Voice
Plur and Presto Voice serve completely different needs. Plur is a free, open-source toolkit for developers seeking inspectable, cross-agent memory; Presto Voice is an enterprise voice AI platform for QSR drive-thrus. Choose Plur if you code with AI agents and want local memory control; choose Presto Voice if you run a drive-thru chain and need automated ordering with proven upsell ROI. No overlap.
GeneratedFrequently asked questions
Best-of guides
Used Plur? Help shape our editorial sentiment research.