How it works · MCP & SDK

A memory network is a groupof people and agents that learn together.

Listen: how memco works

1:06

What one agent learns doing the work, every agent in the network can use, with the trust it has earned. Governed by default through trust and validation rules, with human review where you require it. Nothing you write leaves your network.

For your team For your customers

01System overview

One network. Every agent.

Different tools, different people, one network. Agents read by search, contribute by create and enrich, and signal by feedback. The network ranks, validates and reconciles; agents never coordinate with each other directly.

  • behind your firewallClaude Code

    user · alice

  • behind your firewallCursor

    user · ben

  • behind your firewallWindsurf

    user · carla

  • behind your firewall+ n more

    agents · ci · scripts

  • mcp · sdkmemco instance

    continuous · autonomous

    Hybrid retrieval
    vector + bm25 + trust
    Trust modelling
    bayesian · decays
    Validation rules
    agentic or human review, by config
    Reconciliation
    conflicts resolved · evidence kept
  • read · writeYour network

    private · default
    your team’s knowledge · never leaves the tenant

fig. 01 — agents on the left, one memco instance, your network on the right
tools
search · create_memory · enrich_memory · share_feedback
auth
sso · tenant isolation

02Knowledge domains

Two kinds of work. One connection.

A domain is what a network learns about: what an insight can be about, what its tags mean, and which agents contribute. We ship two, coding and knowledge work. They are fully isolated: a search in one never returns the other, and both are available at the same endpoint.

  • ● coding · agents · claude code · codex · cursorSoftware development.

    API patterns and conventions, error-handling strategies, architectural decisions, dependency-specific behaviours, testing patterns, deployment procedures. Tagged by language, framework and repository.

    • the domain the coding plugins write to
  • ● knowledge · assistants · claude · chatgpt · cowork · grokKnowledge work.

    Process knowledge, policy and compliance, stakeholder and organisational context, domain terminology, analytical frameworks, procedures. The judgement that would otherwise disappear into a chat.

    memco for knowledge work

enterprise customers can define their own domains with us

03Agent interface

Core calls.

Agents talk to memco over MCP, or through the SDK when you own the agent loop. The core read, contribute, feedback cycle is four calls.

  • 01 · read · at task start, on errorssearch(query, context)
    Retrieve relevant insights

    Searches your network. Ranked by a hybrid of vector similarity, full-text match and trust-weighted evidence, not similarity alone.

    • hybrid retrieval · vector + bm25 + trust
  • 02 · write · on discoverycreate_memory(content)
    Store a new insight

    Triggered organically when an agent discovers an undocumented behaviour, a workaround, a convention. No prompt required. Enters the store with an estimated trust score.

    • organic · a side-effect of work
  • 03 · write · on outcomeenrich_memory(id, content)
    Refine an existing insight

    Adds context to an insight the agent has already read: for example, attaching the outcome of a task to the rule that informed it. The primary path for in-session refinement.

    • refinement · keeps insights current in place
  • 04 · signal · after retrievalshare_feedback(id, score)
    Was the insight useful?

    Positive and negative scores both feed the Bayesian trust model. The closing edge of the loop: feedback from one session shapes ranking for every session after it.

    • closes the loop · evidence updates trust

04Trust modelling

Ranked by evidence, not similarity alone.

trust distribution · per insight

Every insight carries a trust score backed by a Bayesian distribution, not a scalar. Each retrieval, each relevance signal, each enrichment updates that distribution.

01 Bayesian, not scalar
Trust is a probability distribution. Each piece of evidence updates the posterior; confidence reflects both the level and the weight of accumulated signal.

02 Decays without signal
Insights that stop being retrieved or stop earning positive feedback gradually lose confidence. Nothing sits on the shelf forever because it once worked.

03 Exploration, not just exploitation
Retrieval samples from the distribution. High-confidence insights surface reliably; lower-evidence insights are selectively included so they can earn signal.

04 Server-side only
Trust is operational. The agent never sees scores and the user never sees them in chat.

 trust →      0        .25       .50       .75        1
 ─────────────┼─────────┼─────────┼─────────┼─────────┼

 new          ▁▂▃▃▃▃▂▁
              wide · low weight

 emerging               ▂▄▆▆▄▂
                        narrowing

 validated                            ▂▆█▆▂
                                      tall · heavy
An insight's confidence is a distribution that narrows with evidence.

05Governance

Governed by default. Human review where you require it.

Trust ranks what has proven useful. Three more mechanisms decide what may be served at all, what must never drift, and what happens when insights disagree.

  • ● validation rules · per network · set by adminsWhat needs a human look

    A network's admins decide what is served straight away and what waits for review, from nothing to everything. Rules can be judged by an LLM against the admin's criteria, so a rule like “anything touching payments” needs no list of keywords.

    • moderation queue · only where a rule says so
  • ● policies · maintained by peopleWhat must not drift

    Knowledge the network treats as always true and never scores. Policies are written and maintained by people, served to every agent, and never decay. They are for the rules that must hold whatever the evidence says.

    • always served · never re-ranked
  • ● reconciliation · continuousWhen insights disagree

    When two insights contradict each other the network resolves the conflict where the evidence allows, flags it where it does not, and keeps the trail either way. Nothing is silently overwritten.

    • resolved or flagged · evidence kept

06Knowledge lifecycle

The loop, end to end.

From the moment an agent contributes, the network takes over. Governed by default through trust and validation rules, with human review where you require it. Operators run continuously and write back into the same store the next search reads from.

  1. 01 · queue · prior · indexIngest

    create or enrich lands, scored and indexed

  2. 02 · rules · review where you require itValidate

    served, or held for a human look

  3. 03 · vector · bm25 · trust · samplingRetrieve

    low-evidence insights get their chance

  4. 04 · synthesise · reconcile · pruneCurate

    operators run continuously

  5. 05 · posterior · re-rankUpdate

    every share_feedback re-scores

feedback re-scores · tomorrow's retrieval is shaped by today's signal

fig. 02 — the loop never breaks

07Networks

Private by default. As many as you need.

Networks are logically isolated in storage and ranked separately at retrieval time. Nothing written in one reaches another unless the networks are arranged so that it does.

  • ● your network · isolation · per tenantOrganisation memory

    Everything your agents contribute lands here, and it never leaves your tenant. Other organisations cannot read it, and memco does not read it. Not shared with other customers, not aggregated, not used for model training.

    • every memory call operates exclusively against it
  • ● hierarchy · parent · childNetworks in a hierarchy

    A child network reads from its parent; writes stay in the child. That is how a team seeds another team, how an integrator seeds its customers, and how roles or product lines get their own memory without losing what is common.

    • read flows down · writes stay local
  • ● always on · boundary · a firewall you controlNothing leaves

    There is no pool your insights drain into. What your agents learn stays inside the network that wrote it, and is never aggregated with anyone else’s. Memory is owned by you.

    • no setting to turn on

Your agents learn. Your models stay the same.

Nothing changes in the weights. The learning lives in the network, so any agent on any model uses it, and it moves with you when you change model, tool or vendor. We have measured learning carried from one model to another.

  • Models
    • Claude
    • GPT
    • Gemini
    • Llama
    • DeepSeek
    • Qwen
  • Agents
    • Claude Code
    • Codex
    • Copilot
    • Claude Desktop
    • ChatGPT
    • Grok
  • IDEs
    • Cursor
    • Windsurf
    • Zed
    • JetBrains
    • VS Code

arXiv preprint — learning carried between two models · τ-bench banking · arXiv:2607.22157

Integration evidence — the stack memco runs against today

09Deployment

Two ways to run it. Same interface.

Both keep your organisation’s memory isolated and expose the same MCP and SDK surface.

  • option a · fully managedManaged, on memco’s cloud.

    Runs on memco’s managed service. Your organisation’s memory is logically isolated: every read and write is scoped to your networks, and role-based access controls who can see and change what. Agents connect over HTTPS from behind your firewall.

    hosting
    memco cloud
    org memory
    logically isolated
    access
    role-based
    time to deploy
    minutes
  • option b · self-hostedOn-premises, in your infra.

    A private memco instance runs entirely behind your firewall. All organisation memory stays on customer-controlled infrastructure. Nothing crosses the firewall.

    hosting
    customer-controlled
    org memory
    stays on-prem
    interface
    identical · mcp & sdk

10Setup path

Running in one session.

Install the plugin in your coding agent, sign in, and the agent has memory. The quick start covers Claude Code, Codex, Cursor and Claude Desktop, and a two-agent tutorial that shows the first contributed and reused insight.

/plugin marketplace add memcoai/marketplace
/plugin install shared-memory@memco
claude code — plugin · marketplace · memcoai/marketplace
Quick start
/quickstart

11Personal memory

Your preferences. Every agent.

Your preferences and strictly personal knowledge, kept out of the shared network and available to every agent you use. Underneath it is a network with its own domain and access rules; to you it is the feature that keeps the team's memory clean.

plugin · personal-memory@memco · or mcp · https://spark.memco.ai/mcp-personal

Personal memory on the team page

12Product FAQ

The technical questions, answered.

Does anything persist on the agent side?

No. Agents talk to memco over MCP or the SDK, and the only thing an agent holds is its session id. Swapping model, IDE or harness changes nothing about the memory.
interface · mcp + sdk

Who decides what gets served?

Governed by default through trust and validation rules, with human review where you require it. Operators run continuously: synthesis, deduplication, reconciliation, safety filtering, pruning. A human look happens only where a validation rule says so.
governance · trust + rules

Can other organisations see our memory?

No. Your network never leaves your tenant. Other organisations cannot read it, memco does not read it, and it is not aggregated or used for model training.
scope · isolated

Which models and tools does it work with?

Any MCP client and any model. The learning lives in the network, below the agent stack, so it survives the next swap of model, IDE or vendor.
portability · mcp-native

What is personal memory?

Your preferences, working style and strictly personal knowledge, kept out of the shared network and available to every agent you use. It is part of memco Shared Memory, private to you, and managed from the dashboard.
scope · personal · private

Does memco work with non-coding agents?

Yes. The knowledge work domain serves assistants such as Claude Desktop, ChatGPT and Grok, with the same trust model, governance and connection.
domains · knowledge work

Can I use memco Shared Memory without MCP?

Yes. The Python and Node.js SDKs give your own code every memory operation, with sessions, typed results and tools for LangChain, the Anthropic SDK and the OpenAI SDK. If you are building the agent, or building memory into a product for your customers, start at For your customers.
interface · sdk

See it on your own repos.

Start a team Try memco free