Aller au contenu

Lab · Mr1000×Growth

Six families of public primitives.

Mr1000×Growth Lab gathers families of applied research and public primitives that irrigate my practice. No social metric is used: no stars, no downloads. Relevance is judged by usage and contract clarity.

Each family distinguishes what can be published (schemas, contracts, scaffolding, notes) from what stays premium in the private practice: enterprise governance, business integrations, dashboards, run, support.

Source code stays private by default; public output is editorialized.

  1. 01Canonical

    Agentic protocols & shared schemas

    Contract layer — jobs, workers, skills, messages, recipes.

    Typed protocols to articulate multi-agent systems without coupling implementations. Versioned schemas, explicit semantics, runtime-agnostic.

    Public layer
    Schemas, types, examples, contract documentation.
    Private / premium layer
    Business profiles, private extensions, SI integrations, migration plans.
  2. 02Active

    Workers & local orchestration

    Runtime layer — daemons, work queues, skill execution.

    Workers designed to attach to a control plane, execute typed skills, surface observable events. Sobriety, resilience, traceability come first.

    Public layer
    Daemon scaffold, adapters, skill scaffolding.
    Private / premium layer
    Hosted control plane, fleet management, run dashboards.
  3. 03Active

    Agentic observability

    Signal layer — typed events, traces, redaction, costs.

    A protocol-first observability contract that makes a multi-agent system inspectable end-to-end: event envelope, session graph, decision ledger, cost meter.

    Public layer
    Event schemas, tracing conventions, instrumentation modules.
    Private / premium layer
    Executive dashboards, alerting, retention, SIEM / data warehouse integrations.
  4. 04Research

    Operational memory

    Memory layer — session, business, doctrine, forget.

    Primitives that distinguish what is forgotten, what is kept and what is published. Memory treated as infrastructure, not as an agent feature.

    Public layer
    Store contracts, forget policy, read / write conventions.
    Private / premium layer
    Enterprise doctrine, granular capabilities, editorial governance.
  5. 05Research

    Source-to-Artifact engine

    Media intelligence layer — transcripts, segments, ideas, exports.

    A chain that turns a source (meeting, video, deck, document) into validatable artifacts. Humans keep editorial decision. Agents prepare.

    Public layer
    Specifications, schemas, local prototypes, Markdown exports.
    Private / premium layer
    Hosted pipelines, CMS / social integrations, rights governance.
  6. 06Prototype

    Model & cost instrumentation

    Measurement layer — wrappers, quality, cost, replay.

    Thin multi-vendor wrappers to tag, compare and trace model calls without coupling agents to a provider.

    Public layer
    Wrappers, schemas, attribution conventions.
    Private / premium layer
    Optimisation plans, cost audits, FinOps integrations.

Publication

Source code stays private by default. Public output happens through essays, notes, specs or demos when the artifact is ready to be cited.