Integrations

One Product Brain. Many doorways.

Same hosted senior PM, same Product Brain, called from wherever your agent already lives. Two doorways are live today — a native MCP server and the plob CLI on npm. The rest are demand-driven.

The distribution channels.

  • Live

    Native MCP server

    Connect from any MCP-aware agent

    Add ProductLobster as a native MCP server in Claude Code, Claude Desktop, Cursor, Windsurf, or Cline. Your agent calls Brain commands directly. The Product Brain runs the work server-side and hands back exactly the result the agent needs.

    plob audit
  • Live

    plob CLI (npm)

    Same API, terminal ergonomics

    Install from npm and call ProductLobster from your terminal or any harness. npm provenance covers supply-chain integrity. Sign in with OAuth, resolve your workspace, and run the same Brain commands the MCP server exposes.

    npm i -g @productlobster/plob
  • Demand-driven

    Claude Code skill

    Drop-in skill for the Claude Code workspace

    A thin Claude Skill that wraps the Brain commands. Install once, call any time. The skill is the doorbell; ProductLobster is the PM who answers. Demand-driven — the MCP server already covers Claude Code today.

    /plob audit
  • Demand-driven

    OpenClaw & Hermes Agent

    Skill bundle for the OpenClaw and Hermes Agent ecosystems

    The same skill, packaged for the OpenClaw and Hermes Agent ecosystems. Same Brain commands, same server-side execution. Demand-driven, not scheduled.

  • Demand-driven

    Printing Press Library

    Read your Product Brain alongside published artifacts

    Printing Press is ProductLobster's published-artifact library: PRDs, decision logs, prototype bundles. The Library makes them addressable from outside the workspace. Demand-driven, not scheduled.

Brain commands you can call today.

Actions run PM work. Reads pull your product's state and your generated documents. Writes capture decisions back to the Brain. All run server-side, against the same compounding memory.

KindCommandWhat it does
Actionplob auditRun a full 5-stage product audit. Returns a run id you can poll.
Actionplob launchRun a full 7-stage launch analysis on a new idea.
Actionplob focused-analysisSingle focused analysis: competitive landscape, customer research, or strategic frame.
Actionplob prototypeGenerate a clickable prototype from agent-supplied context plus Brain state.
Actionplob refineOpen a refinement session on a specific analysis or prototype output.
Readplob context-packPull your product's state into your agent: mental models, summary, locked decisions, run + document index.
Readplob run-status / run-outputPoll a run and pull a completed stage's output. Resolve a pending checkpoint with plob checkpoint-resolve.
Readplob document-list / document-getList generated analysis documents and read the current version of any one.
Readplob workspacesResolve the workspace your credential is bound to. Agents get the slug here, never from the user.
Writeplob observe / ingestCapture a typed decision or signal, or push a working-context bundle the server distills. Through the quarantine gates.
Writeplob document-putWrite your agent's edit back to a document with a version token for safe concurrent edits.

Coming next.

From capturing the decision to querying the why. In active development — join the early-access list.

  • Coming soon
    plob why <decision>

    Trace the reasoning, alternatives, and evidence behind any call.

  • Coming soon
    plob precedent

    Find every decision that depended on a signal, cited a precedent, or superseded an earlier call.

  • Coming soon
    Connectors

    Capture decisions from Linear, your meeting notes, and Slack — the cross-system onramp.

Honesty about timing.

Two doorways are live today: the hosted web app at app.productlobster.ai, and the agentic surface — a native MCP server plus the plob CLI on npm. Both run against the same Product Brain.

The other channels (Claude Code skill, OpenClaw & Hermes Agent, Printing Press Library) are demand-driven, not scheduled — the MCP server already reaches those harnesses. The cross-system connectors (Linear, your meeting notes, Slack) and graph traversal (plob why, plob precedent) are in active development. We don't pin specific dates; join the early-access list and we'll bring you in as they land.

Start in the hosted app today.

The Brain you build now is the same Brain your agent calls from the MCP server and the plob CLI.

Start with your product